Upload
others
View
3
Download
0
Embed Size (px)
Citation preview
Research ArticleRobust Guaranteed Cost Observer Design forSingular Markovian Jump Time-Delay Systems withGenerally Incomplete Transition Probability
Yanbo Li12 Yonggui Kao2 and Jing Xie3
1 School of Mathematical Sciences Guangxi Teachers Education University Guangxi 530001 China2Department of Mathematics Harbin Institute of Technology Weihai 264209 China3 College of Information Science and Engineering Ocean University of China Qingdao 266071 China
Correspondence should be addressed to Yonggui Kao kaoyongguisinacom
Received 29 November 2013 Revised 6 February 2014 Accepted 10 February 2014 Published 24 March 2014
Academic Editor Shen Yin
Copyright copy 2014 Yanbo Li et alThis is an open access article distributed under the Creative Commons Attribution License whichpermits unrestricted use distribution and reproduction in any medium provided the original work is properly cited
This paper is devoted to the investigation of the design of robust guaranteed cost observer for a class of linear singular Markovianjump time-delay systems with generally incomplete transition probability In this singular model each transition rate can becompletely unknown or only its estimate value is known Based on stability theory of stochastic differential equations and linearmatrix inequality (LMI) technique we design an observer to ensure that for all uncertainties the resulting augmented system isregular impulse free and robust stochastically stablewith the proposed guaranteed cost performance Finally a convex optimizationproblem with LMI constraints is formulated to design the suboptimal guaranteed cost filters for linear singular Markovian jumptime-delay systems with generally incomplete transition probability
1 Introduction
Descriptor systems are also referred to as singular systemsimplicit systems generalized state-space systems or semis-tate systems and provide convenient and natural representa-tions in the description of economic systems power systemsrobotics network theory and circuits systems [1] The sta-bility for singular system is more complicated than that fornonsingular systems because not only the asymptotic stabilitybut also the system regularity and impulse elimination areneeded to be addressed [2ndash5]
In practice in many physical systems such as aircraftcontrol solar receiver control power systems manufacturingsystems networked control systems air intake systems andother practical systems abrupt variations may happen intheir structure due to random failures repair of componentssudden environmental disturbances changing subsysteminterconnections or abrupt variations in the operating pointsof a nonlinear plant [6ndash19] Therefore more and moreattention has been paid to the problem of stochastic stabilityand stochastic admissibility for singular Markovian jump
systems (SMJSs) [20ndash30] Long et al [23] derived stochasticadmissibility for a class of singular Markovian jump systemswith mode-dependent time delays Wang and Zhang [27]focused on the asynchronous 119897
2minus119897infinfiltering for discrete-time
stochastic Markov jump systems with randomly occurringsensor nonlinearities However the TRs in the above men-tioned literatures are assumed to be completely known
In practice the TRs in some jumping processes aredifficult to be precisely estimated due to the cost and someother factors Therefore analysis and synthesis problemsfor normal MJSs with incomplete information on transitionprobability have attracted more and more attentions [31ndash49]Xiong and Lam [32] probed robust119867
2control of Markovian
jump systems with uncertain switching probabilities Karanet al [33] considered the stochastic stability robustness forcontinuous-time and discrete-time Markovian jump linearsystems (MJLSs) with upper bounded TRs Zhang andBoukas [34] discussed stability and stabilization for thecontinuous-time MJSs with partly unknown TRs Lin et al[38] considered delay-dependent 119867
infinfiltering for discrete-
time singular Markovian jump systems with time-varying
Hindawi Publishing CorporationAbstract and Applied AnalysisVolume 2014 Article ID 832891 11 pageshttpdxdoiorg1011552014832891
2 Abstract and Applied Analysis
delay and partially unknown transition probabilities Guoand Wang [49] proposed another description for the uncer-tain TRs which is called generally uncertain TRs (GUTRs)
On the other hand state estimation plays an importantrole in systems and control theory signal processing andinformation fusion [50 51] Certainly the most widely usedestimation method is the well-known Kalman filtering [5253] A common feature in the Kalman filtering is that anaccurate model is available In some applications howeverwhen the system is subject to parameter uncertainties theaccurate system model is hard to obtain To overcome thisdifficulty the guaranteed cost filtering approach has beenproposed to ensure the upper bound of guaranteed cost func-tion [54] Robust119867
infinfiltering for uncertainMarkovian jump
systems with mode-dependent time delays was proposed in[55] In [56] guaranteed cost and 119867
infinfiltering for time-
delay systems were presented in terms of LMIs Howeverto the best of our knowledge there are few consideringthe robust guaranteed cost observer for a class of linearsingular Markovian jump time-delay systems with generallyincomplete transition probability which is still an openproblem
In this paper based on LMI method we address thedesign problem of the robust guaranteed cost observer for aclass of uncertain descriptor time-delay systems withMarko-vian jumping parameters and generally uncertain transitionrates The design problem proposed here is to design a mem-oryless observer such that for all uncertainties includinggenerally uncertain transition rates the resulting augmentedsystem is regular impulse-free and robust stochastically sta-ble and satisfies the proposed guaranteed cost performance
2 Problem Formulation
Consider the following descriptor time-delay systems withMarkovian jumping parameters
119864 (119905) = 119860 (119903119905 119905) 119909 (119905) + 119860
119889(119903119905 119905) 119909 (119905 minus 119889)
119910 (119905) = 119862 (119903119905 119905) 119909 (119905) + 119862
119889(119903119905 119905) 119909 (119905 minus 119889)
119909 (119905) = 120593 (119905) forall119905 isin [minus119889 0]
(1)
where 119909(119905) isin 119877119899 and 119910(119905) isin 119877
119903 are the state vector andthe controlled output respectively 119889 represents the state timedelay For convenience the input terms in system (1) havebeen omitted 120593(119905) isin 119871
2[minus119889 0] is a continuous vector-valued
initial function The random parameter 120574(119905) represents acontinuous-time discrete-state Markov process taking valuesin a finite set S = 1 2 119904 and having the transitionprobability matrix Π = [120587
119894119895] 119894 119895 isin 119873 The transition
probability from mode 119894 to mode 119895 is defined by
Pr 119903119905+Δ
= 119895 | 119903119905= 119894 =
120587119894119895Δ + 119900 (Δ) 119894 = 119895
1 + 120587119894119895Δ + 119900 (Δ) 119894 = 119895
(2)
where Δ gt 0 satisfies limΔrarr0
(119900(Δ)Δ) = 0 120587119894119895ge 0 is the
transition probability from mode 119894 to mode 119895 and satisfies
120587119894119894= minus
119904
sum
119895=1119895 = 119894
120587119894119895le 0 (3)
In this paper the transition rates of the jumping processare assumed to be partly available that is some elements inmatrix Λ have been exactly known some have been merelyknown with lower and upper bounds and others may haveno information to use For instance for system (1) withfour operation modes the transition rate matrix might bedescribed by
Λ =
[[[[
[
11+ Δ11
sdot sdot sdot
23+ Δ23
sdot sdot sdot 2119904+ Δ2119904
d
1199042+ Δ1199042
sdot sdot sdot
]]]]
]
(4)
where 119894119895and Δ
119894119895isin [minus120590
119894119895 120590119894119895] (120590119894119895
ge 0) represent theestimate value and estimate error of the uncertain TR 120587
119894119895
respectively where 119894119895and 120590
119894119895are known represents the
complete unknown TR which means that its estimate value119894119895and estimate error bound are unknownFor notational clarity for all 119894 isin S the set 119880
119894
denotes 119880119894
= 119880119894
119896cup 119880119894
119906119896with 119880
119894
119896= 119895 The estimate
value of 120587119894119895is known for 119895 isin S 119880119894
119906119896= 119895 The estimate
value of 120587119894119895is unknown for 119895 isin S Moreover if 119880119894
119896= 0 it is
further described as 119880119894119896= 119896119894
1 119896119894
2 119896
119894
119898 where 119896119894
119898isin N+
represents the 119898th bound-known element with the index119896119894
119898in the 119894th row of matrix Π We assume that the known
estimate values of the TRs are well defined That is
Assumption 1 If 119880119894119896
= S then 119894119895minus 120590119894119895
ge 0 (for all 119895 isin
S 119895 = 119894) 119894119894= minussum
119873
119895=1119895 = 119894119894119895and 120590
119894119894= minussum
119904
119895=1119895 = 119894120590119894119895
Assumption 2 If 119880119894
119896=S and 119894 isin 119880
119894
119896 then
119894119895minus 120590119894119895
ge
0 (for all 119895 isin S 119895 = 119894) 119894119894+ 120590119894119894le 0 and sum
119895isin119880119894
119896
119894119895
Assumption 3 If 119880119894
119896=S and 119894 notin 119880
119894
119896 then
119894119895minus 120590119894119895
ge
0 (for all 119895 isin S)
Remark 4 The above assumption is reasonable since it isthe direct result from the properties of the TRs (eg 120587
119894119895ge
0 (for all 119894 119895 isin S 119895 = 119894) and 120587119894119894= minussum
119904
119895=1119895 = 119894120587119894119895) The above
description about uncertain TRs is more general than eitherthe MJSs model with bounded uncertain TRs or the MJSsmodel with partly uncertain TRs If 119880119894
119906119896= 0 for all 119894 isin S
then generally uncertain TR matrix (4) reduces to boundeduncertain TR matrix (5) as follows
[[[[
[
11+ Δ11
12+ Δ12
sdot sdot sdot 1119904+ Δ1119904
21+ Δ21
22+ Δ22
sdot sdot sdot 2119904+ Δ2119904
d
1199041+ Δ1199041
1199042+ Δ1199042
sdot sdot sdot 119904119904+ Δ119904119904
]]]]
]
(5)
where 119894119895minusΔ119894119895ge 0 (for all 119895 isin S 119895 = 119894)
119894119894= minussum
119904
119895=1119895 = 119894119894119895le
0 and Δ119894119894= sum119904
119895=1119895 = 119894Δ119894119895 if 120590119894119895= 0 for all 119894 isin S for all 119895 isin
Abstract and Applied Analysis 3
119880119894
119896 then generally uncertain TR matrix (4) reduces to partly
uncertain TR matrix (6) as follows
[[[[
[
12058711
sdot sdot sdot
12058723
sdot sdot sdot 1205872119904
d
1205871199042
sdot sdot sdot
]]]]
]
(6)
Our results in this paper can be applicable to the generalMarkovian jump systems with bounded uncertain or partlyuncertain TR matrix
119860(120574(119905) 119905)119860119889(120574(119905) 119905)119862(120574(119905) 119905) and119862
119889(120574(119905) 119905) arematrix
functions of the random jumping process 120574(119905) To simplify thenotion the notation 119860
119894(119905) represents 119860(120574(119905) 119905) when 120574(119905) =
119894 For example 119860119889(120574(119905) 119905) is denoted by 119860
119889119894(119905) and so on
Further for each 120574(119905) = 119894 isin 119873 it is assumed that thematrices 119860
119894(119905) 119860
119889119894(119905) 119862119894(119905) and 119862
119889119894(119905) can be described by
the following form
119860119894(119905) = 119860
119894+ Δ119860119894(119905) 119860
119889119894(119905) = 119860
119889119894+ Δ119860119889119894(119905)
119862119894(119905) = 119862
119894+ Δ119862119894(119905) 119862
119889119894(119905) = 119862
119889119894+ Δ119862119889119894(119905)
(7)
where 119860119894 119860119889119894 119862119894are 119862
119889119894known real coefficient matrices
with appropriate dimensions Time-varying matricesΔ119860119894(119905) Δ119860
119889119894(119905) Δ119862
119894(119905) and Δ119862
119889119894(119905) represent norm-
bounded uncertainties and satisfy
[Δ119860119894(119905) Δ119860
119889119894(119905)
Δ119862119894(119905) Δ119862
119889119894(119905)
] = [1198721119894
1198722119894
]119865119894(119905) [1198731119894 1198732119894] (8)
where119872111989411987221198941198721119894 and119873
2119894are known constant real matri-
ces of appropriate dimensions which represent the structureof uncertainties and 119865
119894(119905) is an unknown matrix function
with Lebesgue measurable elements and satisfies 119865119894(119905)119865119879
119894(119905) le
119868Further for convenience we assume that the system has
the same dimension at each mode and the Markov processis irreducible Consider the following nominal unforceddescriptor time-delay system
119864 (119905) = 119860119894119909 (119905) + 119860
119889119894119909 (119905 minus 119889)
119909 (119905) = 120593 (119905) forall119905 isin [minus119889 0]
(9)
Let 1199090 1199030 and 119909(119905 120593 119903
0) be the initial state initial mode
and the corresponding solution of the system (9) at time 119905respectively
Definition 5 System (9) is said to be stochastically stable iffor all 120593(119905) isin 119871
2[minus119889 0] and initial mode 119903
0isin 119873 there exists
a matrix119872 gt 0 such that
119864int
infin
0
1003817100381710038171003817119909 (119905 120593 1199030)10038171003817100381710038172
119889119905 | 1199030 119909 (119905) = 120593 (119905) 119905 isin [minus119889 0]
le 119909119879
01198721199090
(10)
The following definition can be regarded as an extensionof the definition in [2]
Definition 6 (1) System (9) is said to be regular if det(119904E minus
119860119894) 119894 = 1 2 119904 are not identically zero(2) System (9) is said to be impulse free if deg(det(119904119864 minus
119860119894)) = rank 119864
119894 119894 = 1 2 119904
(3) System (9) is said to be admissible if it is regularimpulse free and stochastically stable
The linearmemoryless observer under consideration is asfollows
119864 119909 (119905) = 1198701119894119909 (119905) + 119870
2119894119910 (119905)
1199090= 0 119903 (0) = 119903
0
(11)
where 119909(119905) isin 119877119899 is the observer state and the constant
matrices1198701119894and119870
2119894are observer parameters to be designed
Denote the error state 119890(119905) = 119909(119905) minus 119909(119905) and theaugmented state vector 119909
119891= [119909119879(119905) 119890119879(119905)]119879
Let 119909(119905) = 119871119890(119905)
represent the output of the error states where 119871 is a knownconstant matrix Define
119860119891119894= [
119860119894
0
119860119894minus 1198701119894minus 11987021198941198621198941198701119894
]
119860119891119889119894
= [119860119889119894
0
119860119889119894minus 1198702119894119862119889119894
0] 119864
119891= [
119864 0
0 119864]
119872119891119894= 1198721198911119894
= [1198721119894
1198721119894minus 11987021198941198722119894
] 119873119891119894= [1198731119894 0]
Δ119860119891119894= 119872119891119894119865119894(119905)119873119891119894 119873
1198911119894= [1198732119894 0]
Δ119860119891119889119894
= 1198721198911119894
119865119894(119905)1198731198911119894
119862119891= [0 119871]
(12)
and combine (1) and (11) then we derive the augmentedsystems as follows
119864119891119891(119905) = (119860
119891119894+ Δ119860119891119894) 119909119891(119905)
+ (119860119891119889119894
+ Δ119860119891119889119894
) 119909119891(119905 minus 119889)
119911 (119905) = 119862119891119909119891(119905)
1199091198910
(119905) = [120593119879(119905) 120593119879(119905)]119879
forall119905 isin [minus119889 0]
(13)
Similar to [5] it is also assumed in this paper that for all 120589 isin[minus119889 0] there exists a scalar ℎ gt 0 such that 119909
119891(119905 + 120589) le
ℎ119909119891(119905)Associated with system (13) is the cost function
J = Eint
infin
0
119911119879(119905) 119911 (119905) 119889119905 (14)
Definition 7 Consider the augmented system (13) if thereexist the observer parameters 119870
1119894 1198702119894and a positive scalar
Jlowast for all uncertainties such that the augmented system(13) is robust stochastically stable and the value of the costfunction (14) satisfies J le Jlowast then Jlowast is said to be arobust guaranteed cost and observer (11) is said to be a robustguaranteed cost observer for system (1) with (4)
Problem 8 (robust guaranteed cost observer problem for aclass of linear singular Markovian jump time-delay systems
4 Abstract and Applied Analysis
with generally incomplete transition probability) Given sys-tem (1) with GUTRMatrix (4) can we determine an observer(11) with parameters 119870
1119894and 119870
2119894such that the observer is a
robust guaranteed cost observer for system (1) with GUTRMatrix (4)
Lemma 9 Given any real number 120576 and any matrix Q thematrix inequality 120576(119876 + 119876
119879) le 1205762119879 + 119876119879
minus1119876119879 holds for any
matrix 119879 gt 0
3 Main Results
Theorem 10 Consider the augmented system (13)with GUTRMatrix (4) and the cost function (14) Then the robust guar-anteed cost observer (11) with parameters 119870
1119894and 119870
2119894can
be designed if there exist matrices 119875119894 1198701119894 and 119870
2119894 119894 =
1 2 119904 and symmetric positive definite matrix Q satisfyingthe following LMIs respectively
Case 1 If 119894 notin 119880119894
119896and 119880
119894
119896= 119896119894
1 119896
119894
119898 there exist a set of
symmetric positive definite matrices 119879119894119895isin R119899times119899 (119894 notin 119880
119894
119896 119895 isin
119880119894
119896) such that
119864119879
119891119875119894= 119875119879
119894119864119891ge 0 (15)
[[
[
Π119894+ 119862119879
119891119862119891
119875119894(119860119891119889119894
+ Δ119860119891119889119894
) 1
(119860119891119889119894
+ Δ119860119891119889119894
)119879
119875119894
minus119876 0
lowast lowast 2
]]
]
lt 0 (16)
119875119894minus 119875119895ge 0 forall119895 isin 119880
119894
119906119896 119895 = 119894 (17)
Case 2 If 119894 isin 119880119894
119896 119880119894119896= 119896119894
1 119896
119894
119898 and 119880119894
119906119896= 0 there exist a
set of symmetric positive definite matrices 119881119894119895119897
isin R119899times119899 (119894 119895 isin
119880119894
119896 119897 isin 119880
119894
119906119896) such that
119864119879
119891119875119894= 119875119879
119894119864119891ge 0 (18)
[[
[
Ω119894+ 119862119879
119891119862119891
119875119894(119860119891119889119894
+ Δ119860119891119889119894
) 1
(119860119891119889119894
+ Δ119860119891119889119894
)119879
119875119894
minus119876 0
lowast lowast 2
]]
]
lt 0 (19)
Case 3 If 119894 isin 119880119894
119896and 119880
119894
119906119896= 0 there exist a set of symmetric
positive definite matrices119882119894119895isin R119899times119899 (119894 119895 isin 119880
119894
119896) such that
119864119879
119891119875119894= 119875119879
119894119864119891ge 0 (20)
[[
[
Δ119894+ 119862119879
119891119862119891
119875119894(119860119891119889119894
+ Δ119860119891119889119894
) 1
(119860119891119889119894
+ Δ119860119891119889119894
)119879
119875119894
minus119876 0
lowast lowast 2
]]
]
lt 0 (21)
where
Π119894= (119860119891119894+ Δ119860119891119894)119879
119875119894+ 119875119894(119860119891119894+ Δ119860119891119894)
+ 119876 + sum
119895isin119880119894
119896
119894119895119864119879
119891(119875119895minus 119875119894) + sum
119895isin119880119894
119896
1
41205902
119894119895119879119894119895
Ω119894= (119860119891119894+ Δ119860119891119894)119879
119875119894+ 119875119894(119860119891119894+ Δ119860119891119894)
+ 119876 + sum
119895isin119880119894
119896
119894119895119864119879
119891(119875119895minus 119875119897) + sum
119895isin119880119894
119896
1
41205902
119894119895119881119894119895119897
Δ119894= (119860119891119894+ Δ119860119891119894)119879
119875119894+ 119875119894(119860119891119894+ Δ119860119891119894)
+ 119876 + sum
119895isinS119895 = 119894
119894119895119864119879
119891(119875119895minus 119875119894) + sum
119895isinS119895 = 119894
1
41205902
119894119895119882119894119895
1= [119864119879
119891(119875119894119896119894
1
minus 119875119894) 119864119879
119891(119875119894119896119894
2
minus 119875119894) 119864
119879
119891(119875119894119896119894
119898
minus 119875119894)]
2= diag minus119879
119894119896119894
1
minus119879119894119896119894
119898
1= [119864119879
119891(119875119896119894
1
minus 119875119897) 119864119879
119891(119875119896119894
2
minus 119875119897) 119864
119879
119891(119875119896119894
119898
minus 119875119897)]
2= diag minus119881
119894119896119894
1119897 minus119881
119894119896119894
119898119897
1= [119864119879
119891(1198751minus 119875119894) 119864
119879
119891(119875119894minus1
minus 119875119894)
119864119879
119891(119875119894+1
minus 119875119894) 119864
119879
119891(119875119904minus 119875119894)]
2= diag minus119882
1198941 minus119882
119894119904
(22)
Proof According to Definition 2 and Theorem 1 in [2] wecan derive from (15)ndash(21) that system (13) is regular andimpulse free Let the mode at time 119905 be 119894 and consider thefollowing Lyapunov function with respect to the augmentedsystem (13)
119881(119909119891(119905) 120574 (119905) = 119894) = 119909
119879
119891(119905) 119864119879
119891119875119894119909119891(119905)
+ int
119905
119905minus119889
119909119879
119891(119904) 119876119909
119891(119904) 119889119905
(23)
where 119876 is the symmetric positive definite matrix to be cho-sen and 119875
119894is a matrix satisfying (15)ndash(21)The weak infinites-
imal operatorL of the stochastic process 120574(119905) 119909119891(119905) 119905 ge 0
is presented by
L119881(119909119891(119905) 120574 (119905) = 119894)
= limΔrarr0
1
Δ[119864119891119881 (119909 (119905 + Δ) 120574 (119905 + Δ)) 119909 (119905) 120574 (119905) = 119894
minus119881 (119909 (119905) 120574 (119905) = 119894) ]
= 119909119879
119891(119905) [
[
(119860119891119894+ Δ119860119891119894)119879
119875119894+ 119875119894(119860119891119894+ Δ119860119891119894)
+
119904
sum
119895=1
120587119894119895119864119879
119891119875119895+ 119876]
]
119909119891(119905)
+ 2119909119879
119891(119905) 119875119894(1198601198911119894
+ Δ1198601198911119894
) 119909119891(119905 minus 119889)
minus 119909119879
119891(119905 minus 119889)119876119909
119891(119905 minus 119889)
(24)
Abstract and Applied Analysis 5
Case 1 (119894 notin 119880119894
119896) Note that in this case sum
119895isin119880119894
119906119896119895 = 119894
120587119894119895
=
minussum119895isinU119894119896119895 = 119894
120587119894119895minus 120587119894119894and 120587
119894119895ge 0 119895 isin 119880
119894
119906119896 119895 = 119894 then from
(24) we have
119909119879
119891(119905) [
[
119904
sum
119895=1
120587119894119895119864119879
119891119875119895]
]
119909119891(119905)
= 119909119879
119891(119905) [
[
sum
119895isin119880119894
119896
120587119894119895119864119879
119891119875119895+ sum
119895isin119880119894
119906119896119895 = 119894
120587119894119895119864119879
119891119875119895+ 120587119894119894119864119879
119891119875119895]
]
119909119891(119905)
= 119909119879
119891(119905) [
[
sum
119895isin119880119894
119896
120587119894119895119864119879
119891119875119895+ (minus120587
119894119894minus sum
119895isin119880119894
119896
120587119894119895)119864119879
119891119875119894
+120587119894119894119864119879
119891119875119894]
]
119909119891(119905)
= 119909119879
119891(119905) 119864119879
119891[
[
sum
119895isin119880119894
119896
120587119894119895(119875119895minus 119875119894)]
]
119909119891(119905)
= 119909119879
119891(119905) 119864119879
119891[
[
sum
119895isin119880119894
119896
119894119895(119875119895minus 119875119894) + sum
119895isin119880119894
119896
Δ119894119895(119875119895minus 119875119894)]
]
119909119891(119905)
(25)
On the other hand in view of Lemma 9 we have
sum
119895isin119880119894
119896
Δ119894119895119864119879
119891(119875119895minus 119875119894)
= sum
119895isin119880119894
119896
[1
2Δ119894119895119864119879
119891(119875119895minus 119875119894) +
1
2Δ119894119895119864119879
119891(119875119895minus 119875119894)]
le sum
119895isin119880119894
119896
[(1
2Δ119894119895)
2
119879119894119895+ 119864119879
119891(119875119895minus 119875119894) 119879minus1
119894119895(119875119895minus 119875119894) 119864119891]
le sum
119895isin119880119894
119896
[1
41205902
119894119895119879119894119895+ 119864119879
119891(119875119895minus 119875119894) 119879minus1
119894119895(119875119895minus 119875119894) 119864119891]
(26)
Case 2 (119894 isin 119880119894
119896and119880119894
119906119896= 0) Because of119880119894
119896= 119896119894
1 119896
119894
119898 and
119880119894
119906119896= 119906119894
1 119906
119894
119904minus119898 there must be 119897 isin 119880
119894
119906119896so that 119864119879
119891119875119897ge
119864119879
119891119875119895(for all 119895 isin 119880
119894
119906119896)
119909119879
119891(119905) [
[
119904
sum
119895=1
120587119894119895119864119879
119891119875119895]
]
119909119891(119905)
le 119909119879
119891(119905) [
[
sum
119895isin119880119894
119896
120587119894119895119864119879
119891119875119895minus ( sum
119895isin119880119894
119896119895 = 119894
120587119894119895)119864119879
119891119875119897]
]
119909119891(119905)
= 119909119879
119891(119905) 119864119879
119891[
[
sum
119895isin119880119894
119896
120587119894119895(119875119895minus 119875119897)]
]
119909119891(119905)
= 119909119879
119891(119905) 119864119879
119891[
[
sum
119895isin119880119894
119896
119894119895(119875119895minus 119875119897) + sum
119895isin119880119894
119896
Δ119894119895(119875119895minus 119875119897)]
]
119909119891(119905)
(27)
By using Lemma 9 we have
sum
119895isin119880119894
119896
Δ119894119895119864119879
119891(119875119895minus 119875119897)
= sum
119895isin119880119894
119896
[1
2Δ119894119895119864119879
119891(119875119895minus 119875119897) +
1
2Δ119894119895119864119879
119891(119875119895minus 119875119897)]
le sum
119895isin119880119894
119896
[(1
2Δ119894119895)
2
119881119894119895119897+ 119864119879
119891(119875119895minus 119875119897)119881minus1
119894119895119897(119875119895minus 119875119897) 119864119879
119891]
le sum
119895isin119880119894
119896
[1
41205902
119894119895119881119894119895119897+ 119864119879
119891(119875119895minus 119875119897)119881minus1
119894119895119897(119875119895minus 119875119897) 119864119879
119891]
(28)
Case 3 (119894 isin 119880119894
119896and 119880
119894
119906119896= 0) Consider
119909119879
119891(119905) [
[
119904
sum
119895=1
120587119894119895119864119879
119891119875119895]
]
119909119891(119905)
= 119909119879
119891(119905) 119864119879
119891[
[
119904
sum
119895=1119895 = 119894
120587119894119895(119875119895minus 119875119894)]
]
119909119891(119905)
= 119909119879
119891(119905) 119864119879
119891[
[
119904
sum
119895=1119895 = 119894
119894119895(119875119895minus 119875119894)
+
119904
sum
119895=1119895 = 119894
Δ119894119895(119875119895minus 119875119894)]
]
119909119891(119905)
(29)
Case 1 Substituting (25) and (26) into (24) it results in
L119881 le Λ119879(119905) Φ (119894) Λ (119905) (30)
where Λ119879(119905) = [119909119879
119891(119905) 119909119879
119891(119905 minus 119889)] and
Φ119894=[[
[
Π119894+ 119862119879
119891119862119891
119875119894(119860119891119889119894
+ Δ119860119891119889119894
) 1
(119860119891119889119894
+ Δ119860119891119889119894
)119879
119875119894
minus119876 0
lowast lowast 2
]]
]
(31)
Case 2 Substituting (27) and (28) into (24) it results in
L119881 le Λ119879(119905) Ψ (119894) Λ (119905) (32)
where Λ119879(119905) = [119909119879
119891(119905) 119909119879
119891(119905 minus 119889)] and
Ψ119894=[[
[
Ω119894+ 119862119879
119891119862119891
119875119894(119860119891119889119894
+ Δ119860119891119889119894
) 1
(119860119891119889119894
+ Δ119860119891119889119894
)119879
119875119894
minus119876 0
lowast lowast 2
]]
]
(33)
6 Abstract and Applied Analysis
Case 3 Substituting (29) into (24) we get
L119881 le Λ119879(119905) Γ (119894) Λ (119905) (34)
where Λ119879(119905) = [119909119879
119891(119905) 119909119879
119891(119905 minus 119889)] and
Γ119894=[[
[
Δ119894+ 119862119879
119891119862119891
119875119894(119860119891119889119894
+ Δ119860119891119889119894
) 1
(119860119891119889119894
+ Δ119860119891119889119894
)119879
119875119894
minus119876 0
lowast lowast 2
]]
]
(35)
Similar to [5] usingDynkinrsquos formula we drive for each 119894 isin 119873
lim119879rarrinfin
Eint
119879
0
119909119879
119891(119905) 119909119891(119905) 119889119905 | 120593
119891 1205740= 119894 le 119909
119879
11989101198721199091198910 (36)
By Definition 5 it is easy to see that the augmented system(13) is stochastically stable Furthermore from (16) (19) and(21) we have
L119881 le minus119909119879
119891(119905) 119862119879
119891119862119891119909119891(119905) lt 0 (37)
On the other hand we have
J = Eint
infin
0
119909119879
119891(119905) 119862119879
119891119862119891119909119891(119905) 119889119905 lt minusint
infin
0
L119881119889119905
= minus E lim119905rarrinfin
119881 (119909 (119905) 120574 (119905)) + 119881 (1199090 1205740)
(38)
As the augmented system (13) is stochastically stable itfollows from (38) that 119869 lt 119881(119909
1198910 1199030) From Definition 7 it
is concluded that a robust guaranteed cost for the augmentedsystem (13) can be given by 119869
lowast= 119909119879
1198910(119905)119864119879
1198911199030119875(1199030)1199091198910
+
int0
minus119889119909119879
119891(119905)119876119909
119891(119905)119889119905
In the following based on the above sufficient conditionthe design of robust guaranteed cost observers can be turnedinto the solvability of a system of LMIs
Theorem 11 Consider system (13)with GUTRMatrix (4) andthe cost function (14) If there exist matrices 119884
1119894and 119884
2119894 119894 =
1 2 119904 positive scalars 120576119894 119894 = 1 2 119904 symmetric positive
definite matrix Q and the full rank matrices 1198752119894 and matrices
119875119894= diag(119875
1119894 1198752119894) 119894 = 1 2 119904 satisfying the following
LMIs respectively
Case 1 If 119894 notin 119880119894
119896and 119880
119894
119896= 119896119894
1 119896
119894
119898 a set of positive
definite matrices 119879119894119895isin R119899times119899 (119894 notin 119880
119894
119896 119895 isin 119880
119894
119896) exist such that
119864119879
119891119875119894= 119875119879
119894119864119891ge 0 (39)
[[[[[[[[[
[
1206011119894
1206012119894
1198731
1206013119894
120601119879
2119894minus119876 0 0
119873119879
10 119873
20
120601119879
31198940 0 minus120576
119894119868
]]]]]]]]]
]
lt 0 (40)
119875119894minus 119875119895ge 0 forall119895 isin 119880
119894
119906119896 119895 = 119894 (41)
Case 2 If 119894 isin 119880119894
119896(119880119894
119896= 119896119894
1 119896
119894
119898) and 119880
119894
119906119896= 0 a set of
positive definite matrices119881119894119895119897isin R119899times119899 (119894 119895 isin 119880
119894
119896 119897 isin 119880
119894
119906119896) exist
such that
119864119879
119891119875119894= 119875119879
119894119864119891ge 0 (42)
[[[[[[[[[
[
1205931119894
1205932119894
1198721
1205933119894
120593119879
2119894minus119876 0 0
119872119879
10 119872
20
120593119879
31198940 0 minus120576
119894119868
]]]]]]]]]
]
lt 0 (43)
Case 3 If 119894 isin 119880119894
119896and 119880
119894
119906119896= 0 a set of positive definite
matrices119882119894119895isin R119899times119899 (119894 119895 isin 119880
119894
119896) exist such that
119864119879
119891119894119875119894= 119875119879
119894119864119891119894ge 0 (44)
[[[[[[[[[
[
1205951119894
1205952119894
1198711
1205953119894
120595119879
2119894minus119876 0 0
119871119879
10 1198712
0
120595119879
31198940 0 minus120576
119894119868
]]]]]]]]]
]
lt 0 (45)
where
1206011119894= 1205931119894= 1205951119894
= [1198751119894119860119894+ 119860119879
1198941198751119894
119860119879
1198941198752119894minus 119884119879
1119894minus 119862119879
119894119884119879
2119894
1198752119894119860119894minus 1198841119894minus 1198842119894119862119894
119884119879
1119894+ 1198841119894
]
+ 119876 + 119862119879
119891119862119891+ sum
119895isin119880119894
119896
119894119895119864119879
119891(119875119895minus 119875119894) + sum
119895isin119880119894
119896
1
41205902
119894119895119879119894119895
1206012119894= 1205932119894= 1205952119894= [
1198751119894119860 i 0
1198752119894119860119894minus 1198841119894minus 11988421198941198621198940]
1206013119894= 1205933119894= 1205953119894= [
11987511198941198721119894
11987521198941198721119894minus 11988411198941198721119894minus 11988421198941198722119894
]
1198731= [119864119879
119891(119875119896119894
1
minus 119875119894) 119864119879
119891(119875119896119894
2
minus 119875119894) 119864
119879
119891(119875119896119894
119898
minus 119875119894)]
1198732= diag minus119879
119894119896119894
1
minus119879119894119896119894
119898
1198721= [119864119879
119891(119875119896119894
1
minus 119875119894) 119864119879
119891(119875119896119894
2
minus 119875119894) 119864
119879
119891(119875119896119894
119898
minus 119875119894)]
1198722= diag minus119881
119894119896119894
1119897 minus119881
119894119896119894
119898119897
1198711= [119864119879
119891(1198751minus 119875119894) 119864
119879
119891(119875119894minus1
minus 119875119894)
119864119879
119891(119875119894+1
minus 119875119894) 119864
119879
119891(119875119904minus 119875119894)]
1198712= diag minus119882
1198941 minus119882
119894119904
(46)
Abstract and Applied Analysis 7
Then a suitable robust guaranteed cost observer in theform of (11) has parameters as follows
1198701119894= 119875minus1
11198941198841119894 119870
2119894= 119875minus1
21198941198842119894
(47)
and 119869lowast is a robust guaranteed cost for system (13) with GUTRMatrix (4)
Proof Define
1198601
119894=
[[[[
[
119860119879
119891119894119875119894+ 119875119894119860119891119894+ 119876 + sum
119895isin119880119894
119896
119894119895(119875119895minus 119875119894) + sum
119895isin119880119894
119896
1
41205902
119894119895119879119894119895+ 119862119879
119891119862119891
119875119894119860119891119889119894
1198731
119860119879
119891119889119894119875119894
minus119876 0
lowast lowast 1198732
]]]]
]
(48)
1198602
119894=
[[[[
[
119860119879
119891119894119875119894+ 119875119894119860119891119894+ 119876 + sum
119895isin119880119894
119896
119894119895(119875119895minus 119875119894) + sum
119895isin119880119894
119896
1
41205902
119894119895119881119894119895119897+ 119862119879
119891119862119891
119875119894119860119891119889119894
1198721
119860119879
119891119889119894119875119894
minus119876 0
lowast lowast 1198722
]]]]
]
(49)
1198603
119894=
[[[[
[
119860119879
119891119894119875119894+ 119875119894119860119891119894+ 119876 + sum
119895isin119880119894
119896
119894119895(119875119895minus 119875119894) + sum
119895isin119880119894
119896
1
41205902
119894119895119882119894119895+ 119862119879
119891119862119891
119875119894119860119891119889119894
1198711
119860119879
119891119889119894119875119894
minus119876 0
lowast lowast 1198712
]]]]
]
lt 0 (50)
Then (16) is equivalent to
1198601
119894+ [
[
119875119894119872119891119894
0
0
]
]
119865119894[119873119891119894 1198731198911119894 0]
+ [119873119891119894 1198651198911119894 0]119879
119865119879
119894[
[
119875119894119872119891119894
0
0
]
]
119879
lt 0
(51)
Then (19) is equivalent to
1198602
119894+ [
[
119875119894119872119891119894
0
0
]
]
119865119894[119873119891119894 1198731198911119894 0]
+ [119873119891119894 1198651198911119894 0]119879
119865119879
119894[
[
119875119894119872119891119894
0
0
]
]
119879
lt 0
(52)
Then (21) is equivalent to
1198603
119894+ [
[
119875119894119872119891119894
0
0
]
]
119865119894[119873119891119894 1198731198911119894 0]
+ [119873119891119894 1198651198911119894 0]119879
119865119879
119894[
[
119875119894119872119891119894
0
0
]
]
119879
lt 0
(53)
By applying Lemma 24 in [57] (50) (51) and (52) hold forall uncertainties119865i satisfying119865
119879
119894119865119894lt 119868 if and only if there exist
positive scalars 120576119894 119894 = 1 2 119904 such that
1198601
119894+ 120576minus1
119894[
[
119875119894119872119891119894
0
0
]
]
[
[
119875119894119872119891119894
0
0
]
]
119879
+ 120576119894[119873119891119894 1198651198911119894 0]
119879
[119873119891119894 1198651198911119894 0] lt 0
1198602
119894+ 120576minus1
119894[
[
119875119894119872119891119894
0
0
]
]
[
[
119875119894119872119891119894
0
0
]
]
119879
+ 120576119894[119873119891119894 1198651198911119894 0]
119879
[119873119891119894 1198651198911119894 0] lt 0
1198603
119894+ 120576minus1
119894[
[
119875119894119872119891119894
0
0
]
]
[
[
119875119894119872119891119894
0
0
]
]
119879
+ 120576119894[119873119891119894 1198651198911119894 0]
119879
[119873119891119894 1198651198911119894 0] lt 0
(54)
Let 119875119894= diag(119875
1119894 1198752119894) and using (47) we can conclude from
Schur complement results that the above matrix inequalitiesare equivalent to the coupled LMIs (40) (43) and (45)It further follows from Theorem 10 that 119869
lowast is a robustguaranteed cost for system (13) with (4)
Remark 12 The solution of LMIs (39)ndash(45) parameterizesthe set of the proposed robust guaranteed cost observersThis parameterized representation can be used to design theguaranteed cost observer with some additional performanceconstraints By applying the methods in [14] the suboptimal
8 Abstract and Applied Analysis
guaranteed cost observer can be determined by solving acertain optimization problem This is the following theorem
Theorem 13 Consider system (13)with GUTRMatrix (4) andthe cost function (14) and suppose that the initial conditions 119903
0
and 1199091198910
are known if the following optimization problem
min1198761198751119894 1198752119894 120576119894 1198841119894 and 1198842119894
119869lowast
st LMIs (39) ndash (45)(55)
has a solution 119876 1198751119894 1198752119894 120576119894 1198841119894 and 119884
2119894 119894 = 1 2 119904
then the observer (11) is a suboptimal guaranteed costobserver for system (1) where 119869
lowast= 119909119879
1198910119864119879
1198911199030119875(1199030)1199091198910
+
tr(int0minus119889
1199091198910(119905)1199091198910(119905)119909119879
1198910119889119905119876)
Proof It follows from Theorem 11 that the observer (11)constructed in terms of the solution 119876 119875
1119894 1198752119894 120576119894 1198841119894 and
1198842119894 119894 = 1 2 119904 is a robust guaranteed cost observer By
noting that
int
0
minus119889
119909119879
1198910(119905) 119876119909
1198910(119905) 119889119905 = int
0
minus119889
tr (1199091198791198910
(119905) 1198761199091198910
(119905)) 119889119905
= tr(int0
minus119889
119909119879
1198910(119905) 1199091198910
(119905) 119889119905119876)
(56)
it follows that the suboptimal guaranteed cost observerproblem is turned into the minimization problem (55)
Remark 14 Theorem 13 gives the suboptimal guaranteed costobserver conditions of a class of linear Markovian jump-ing time-delay systems with generally incomplete transitionprobability and LMI constraints which can be easily solvedby the LMI toolbox in MATLAB
4 Numerical Example
In this section a numerical example is presented todemonstrate the effectiveness of the method mentioned inTheorem 11 Consider a 2-dimensional system (1) with 3Markovian switching modes In this numerical example thesingular system matrix is set as 119864 = [ 1 0
0 0] and the 3-mode
transition rate matrix is Λ = [minus32
2
15 21 minus36
] where Δ11 Δ31
isin
[minus015 015] Δ23 Δ33
isin [minus012 012] and Δ32
isin [minus01 01]The other system matrices are as follows
For mode 119894 = 1 there are
1198601= [
minus32 065
1 02] 119860
1198891= [
02 05
1 minus068]
1198621= [
[
12 065
minus65 19
minus021 minus18
]
]
1198621198891
= [
[
minus36 minus105
21 096
021 minus086
]
]
11987211
= [minus02
08] 119872
21= [
[
025
0875
minus2
]
]
11987311
= [minus12 31] 11987321
= [minus069 minus42]
(57)
For mode 119894 = 2 there are
1198602= [
minus1 6
2 minus36] 119860
1198892= [
minus31 minus16
3 075]
1198622= [
[
9 minus25
035 minus2
36 minus18
]
]
1198621198892
= [
[
089 minus6
minus12 09
minus24 6
]
]
11987212
= [23
minus4] 119872
22= [
[
075
minus36
25
]
]
11987312
= [minus72 minus6] 11987322
= [1 2]
(58)
For mode 119894 = 3 there are
1198603= [
minus106 29
minus03 36] 119860
1198893= [
minus56 minus12
minus3 45]
1198623= [
[
minus3 minus036
015 minus18
09 minus5
]
]
1198621198893
= [
[
minus165 5
minus12 265
minus098 minus56
]
]
11987213
= [minus82
minus03] 119872
23= [
[
minus052
25
minus36
]
]
11987313
= [105 minus5] 11987323
= [minus72 minus126]
(59)
Then we set the error state matrix 119871 = [ minus45 062 minus6
] and thepositive scalars in Theorem 11 are 120576
1= 02 120576
2= 015 120576
3=
032 According to the definitions of augmented statematricesin (12) we can easily obtain the following parameter matricesin Theorem 11 by MATLAB
11988411
= [minus84521006 00127
00127 84509001]
11988421
= [002 01291 01435
minus23520 minus04080 minus08106]
11988412
= [minus170991 269626
269626 minus246750]
11988422
= [minus200744 minus132941 529388
216893 180120 minus506693]
11988413
= [minus6751329 224456
224456 minus8976976]
11988423
= [minus136021 minus1461726 540500
minus34324 minus195125 minus103068]
1198751=[[[
[
63029 0 0 0
0 48620 0 0
0 0 22914 0
0 0 0 03169
]]]
]
Abstract and Applied Analysis 9
1198752=[[[
[
08914 0 0 0
0 12505 0 0
0 0 73629 0
0 0 0 30056
]]]
]
1198753=[[[
[
30265 0 0 0
0 02156 0 0
0 0 08965 0
0 0 0 10002
]]]
]
119876 =[[[
[
05000 0 0 0
0 05001 0 0
0 0 05001 0
0 0 0 05001
]]]
]
11987911
=[[[
[
34173214 minus8707765 0 0
minus8707765 4167216 0 0
0 0 22263598 minus3207456
0 0 minus3207456 12263101
]]]
]
11987923
=[[[
[
37753231 minus27999330 0 0
minus27999330 28107685 0 0
0 0 106907366 minus107432750
0 0 minus107432750 108555053
]]]
]
11987931
=[[[
[
9518504 minus5399245 0 0
minus5399245 8962029 0 0
0 0 14773012 minus2077540
0 0 minus2077540 14791256
]]]
]
11987932
=[[[
[
21617695 minus12094164 0 0
minus12094164 20371205 0 0
0 0 14786 minus09283
0 0 minus09283 14794
]]]
]
11987933
=[[[
[
14939313 minus8398780 0 0
minus8398780 14073689 0 0
0 0 1478123 minus1336452
0 0 minus1336452 2459347
]]]
]
11988111
=[[[
[
16650 0 0 0
0 16650 0 0
0 0 16650 0
0 0 0 16650
]]]
]
11988231
=[[[
[
15426 0 0 0
0 16650 0 0
0 0 16662 0
0 0 0 16650
]]]
]
11988232
=[[[
[
15428 0 0 0
0 16650 0 0
0 0 16622 0
0 0 0 16650
]]]
]
(60)
Therefore we can design a linear memoryless observer as(11) with the constant matrices
11987011
= 119875minus1
1111988411
= [minus13409860 00020
00026 17381530]
11987021
= 119875minus1
2111988421
= [00087 00563 00626
minus74219 minus12875 minus25579]
11987012
= 119875minus1
1211988412
= [minus191823 302475
215615 minus197321]
11987022
= 119875minus1
2211988422
= [minus27264 minus18056 71899
72163 59928 minus168583]
11987013
= 119875minus1
1311988413
= [minus2231402 74186
1041076 minus41637180]
11987023
= 119875minus1
2311988423
= [minus151724 minus1630481 602900
minus34317 minus195086 minus103047]
(61)
Finally the observer (11) with the above parameter matri-ces for this numerical example is a suboptimal guaranteedcost observer byTheorems 11 and 13
5 Conclusions
In this paper the robust guaranteed cost observer problemfor a class of uncertain descriptor time-delay systems withMarkovian jumping parameters and generally uncertaintransition rates is studied by using LMI method In thisGUTR singular model each transition rate can be com-pletely unknown or only its estimate value is known Theparameterrsquos uncertainty is time varying and is assumed tobe norm-bounded Memoryless guaranteed cost observersare designed in terms of a set of linear coupled matrixinequalities The suboptimal guaranteed cost observer isdesigned by solving a certain optimization problem Ourresults can be applicable to the general Markovian jumpsystems with bounded uncertain or partly uncertain TRmatrix
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This research is supported by the National NaturalScience Foundations of China (10701020 1097102260973048 61272077 60974025 60673101 and 60939003)NCET-08-0755 National 863 Plan Project (2008AA04Z401 2009AA043404) SDPW IMZQWH010016the Natural Science Foundation of Shandong Province (noZR2012FM006) the Scientific and Technological Projectof Shandong Province (no 2007GG3WZ04016) and theNatural Science Foundation of Guangxi Autonomous Region(no 2012GXNSFBA053003)
10 Abstract and Applied Analysis
References
[1] L Dai Singular Control Systems Springer Berlin Germany1989
[2] S Xu P V Dooren R Stefan and J Lam ldquoRobust stability andstabilization for singular systemswith state delay and parameteruncertaintyrdquo IEEE Transactions on Automatic Control vol 47no 7 pp 1122ndash1128 2002
[3] M Fu and G Duan ldquoRobust guaranteed cost observer foruncertain descriptor time-delay systems withMarkovian jump-ing parametersrdquo Acta Automatica Sinica vol 31 pp 479ndash4832005
[4] Y Q Xia E-K Boukas P Shi and J H Zhang ldquoStabilityand stabilization of continuous-time singular hybrid systemsrdquoAutomatica vol 45 no 6 pp 1504ndash1509 2009
[5] S Xu and J Lam Robust Control and Filtering of SingularSystems Springer Berlin Germany 2006
[6] Y-Y Cao and J Lam ldquoRobust 119867infin
control of uncertainMarkovian jump systems with time-delayrdquo IEEE Transactionson Automatic Control vol 45 no 1 pp 77ndash83 2000
[7] Y Li and Y Kao ldquoStability of stochastic reaction-diffusion sys-tems with Markovian switching and impulsive perturbationsrdquoMathematical Problems in Engineering vol 2012 Article ID429568 13 pages 2012
[8] X Mao and C Yuan Stochastic Differential Equations withMarkovian Switching Imperial College Press London UK2006
[9] E-K Boukas Stochastic Switching Systems Analysis and DesignBirkhauser Basel Switzerland 2005
[10] Y Kao C Wang and L Zhang ldquoDelay-dependent exponentialstability of impulsive markovian jumping Cohen-Grossbergneural networks with reaction-diffusion and mixed delaysrdquoNeural Processing Letters vol 38 no 3 pp 321ndash346 2013
[11] Y-G Kao J-F Guo C-H Wang and X-Q Sun ldquoDelay-dependent robust exponential stability of Markovian jump-ing reaction-diffusion Cohen-Grossberg neural networks withmixed delaysrdquo Journal of the Franklin Institute vol 349 no 6pp 1972ndash1988 2012
[12] Y Kao C Wang F Zha and H Cao ldquoStability in mean ofpartial variables for stochastic reactionmdashdiffusion systems withMarkovian switchingrdquo Journal of the Franklin Institute vol 351no 1 pp 500ndash512 2014
[13] N Kumaresan and P Balasubramaniam ldquoOptimal control forstochastic nonlinear singular system using neural networksrdquoComputers amp Mathematics with Applications vol 56 no 9 pp2145ndash2154 2008
[14] L Yu and J Chu ldquoAn LMI approach to guaranteed cost controlof linear uncertain time-delay systemsrdquoAutomatica vol 35 no6 pp 1155ndash1159 1999
[15] S Yin H Luo and S Ding ldquoReal-time implementation of fault-tolerant control systems with performance optimizationrdquo IEEETransactions on Industrial Electronics vol 64 no 5 pp 2402ndash2411 2014
[16] S Yin G Wang and H Karimi ldquoData-driven design of robustfault detection system for wind turbinesrdquoMechatronics 2013
[17] S Yin S X Ding A H A Sari and H Hao ldquoData-drivenmonitoring for stochastic systems and its application on batchprocessrdquo International Journal of Systems Science vol 44 no 7pp 1366ndash1376 2013
[18] S Yin S Ding A Haghani H Hao and P Zhang ldquoA com-parison study of basic datadriven fault diagnosis and process
monitoring methods on the benchmark Tennessee Eastmanprocessrdquo Journal of Process Control vol 22 no 9 pp 1567ndash15812012
[19] Y G Kao and C H Wang ldquoGlobal stability analysis forstochastic coupled reaction-diffusion systems on networksrdquoNonlinear Analysis Real World Applications vol 14 no 3 pp1457ndash1465 2013
[20] E K Boukas ldquoStabilization of stochastic singular nonlinearhybrid systemsrdquo Nonlinear Analysis Theory Methods amp Appli-cations vol 64 no 2 pp 217ndash228 2006
[21] E-K Boukas Control of Singular Systems with Random AbruptChanges Springer Berlin Germany 2008
[22] Y Ding H Zhu S Zhong and Y Zeng ldquoExponential mean-square stability of time-delay singular systems with Markovianswitching and nonlinear perturbationsrdquo Applied Mathematicsand Computation vol 219 no 4 pp 2350ndash2359 2012
[23] S Long S Zhong and Z Liu ldquoStochastic admissibility fora class of singular Markovian jump systems with mode-dependent time delaysrdquoAppliedMathematics and Computationvol 219 no 8 pp 4106ndash4117 2012
[24] Z Wu P Shi H Su and J Chu ldquoAsynchronous 1198682minus 119868infin
filtering for discrete-time stochastic Markov jump systems withrandomly occurred sensor nonlinearitiesrdquo Automatica vol 50no 1 pp 180ndash186 2014
[25] Z Wu P Shi H Su and J Chu ldquoStochastic synchronizationof Markovian jump neural networks with time-varying delayusing sampled-datardquo IEEE Transactions on Cybernetics vol 43no 6 pp 1796ndash1806 2013
[26] S Ma and E-K Boukas ldquoGuaranteed cost control of uncertaindiscrete-time singular Markov jump systems with indefinitequadratic costrdquo International Journal of Robust and NonlinearControl vol 21 no 9 pp 1031ndash1045 2011
[27] G Wang and Q Zhang ldquoRobust control of uncertain singularstochastic systems with Markovian switching via proportional-derivative state feedbackrdquo IET Control Theory amp Applicationsvol 6 no 8 pp 1089ndash1096 2012
[28] J Wang H Wang A Xue and R Lu ldquoDelay-dependent 119867infin
control for singular Markovian jump systems with time delayrdquoNonlinear Analysis Hybrid Systems vol 8 pp 1ndash12 2013
[29] L Wu X Su and P Shi ldquoSliding mode control with bounded1198712gain performance of Markovian jump singular time-delay
systemsrdquo Automatica vol 48 no 8 pp 1929ndash1933 2012[30] Z-GWu J H ParkH Su and J Chu ldquoStochastic stability anal-
ysis for discrete-time singular Markov jump systems with time-varying delay and piecewise-constant transition probabilitiesrdquoJournal of the Franklin Institute vol 349 no 9 pp 2889ndash29022012
[31] J Xiong J Lam H Gao and D W C Ho ldquoOn robust stabi-lization of Markovian jump systems with uncertain switchingprobabilitiesrdquo Automatica vol 41 no 5 pp 897ndash903 2005
[32] J Xiong and J Lam ldquoRobust 1198672control of Markovian jump
systems with uncertain switching probabilitiesrdquo InternationalJournal of Systems Science vol 40 no 3 pp 255ndash265 2009
[33] M Karan P Shi and C Y Kaya ldquoTransition probabilitybounds for the stochastic stability robustness of continuous-and discrete-timeMarkovian jump linear systemsrdquoAutomaticavol 42 no 12 pp 2159ndash2168 2006
[34] L Zhang and E-K Boukas ldquoStability and stabilization ofMarkovian jump linear systems with partly unknown transitionprobabilitiesrdquo Automatica vol 45 no 2 pp 463ndash468 2009
Abstract and Applied Analysis 11
[35] B Du J Lam Y Zou and Z Shu ldquoStability and stabilization forMarkovian jump time-delay systems with partially unknowntransition ratesrdquo IEEE Transactions on Circuits and Systems IRegular Papers vol 60 no 2 pp 341ndash351 2013
[36] L Zhang and J Lam ldquoNecessary and sufficient conditionsfor analysis and synthesis of Markov jump linear systemswith incomplete transition descriptionsrdquo IEEE Transactions onAutomatic Control vol 55 no 7 pp 1695ndash1701 2010
[37] Y Guo and F Zhu ldquoNew results on stability and stabilizationof Markovian jump systems with partly known transitionprobabilitiesrdquoMathematical Problems in Engineering vol 2012Article ID 869842 11 pages 2012
[38] J Lin S Fei and J Shen ldquoDelay-dependent 119867infin
filtering fordiscrete-time singular Markovian jump systems with time-varying delay and partially unknown transition probabilitiesrdquoSignal Processing vol 91 no 2 pp 277ndash289 2011
[39] J Tian Y Li J Zhao and S Zhong ldquoDelay-dependent stochasticstability criteria for Markovian jumping neural networks withmode-dependent time-varying delays and partially knowntransition ratesrdquo Applied Mathematics and Computation vol218 no 9 pp 5769ndash5781 2012
[40] Y Wei J Qiu H R Karimi and M Wang ldquoA new design 119867infin
filtering for contrinuous-time Markovian jump systems withtime-varying delay and partially accessible mode informationrdquoSignal Processing vol 93 pp 2392ndash2407 2013
[41] L Zhang E-K Boukas and J Lam ldquoAnalysis and synthesisof Markov jump linear systems with time-varying delays andpartially known transition probabilitiesrdquo IEEE Transactions onAutomatic Control vol 53 no 10 pp 2458ndash2464 2008
[42] L Zhang and E-K Boukas ldquoMode-dependent 119867infin
filteringfor discrete-time Markovian jump linear systems with partlyunknown transition probabilitiesrdquo Automatica vol 45 no 6pp 1462ndash1467 2009
[43] L Zhang and E-K Boukas ldquoMode-dependent 119867infin
controlfor discrete-time Markovian jump linear systems with partlyunknown transition probabilitiesrdquo International Journal ofRobust and Nonlinear Control vol 19 no 8 pp 868ndash883 2009
[44] Y Zhang Y He M Wu and J Zhang ldquoStabilization forMarkovian jump systems with partial information on transi-tion probability based on free-connection weighting matricesrdquoAutomatica vol 47 no 1 pp 79ndash84 2011
[45] G Zong D Yang L Hou and Q Wang ldquoRobust finite-time119867infincontrol for Markovian jump systems with partially known
transition probabilitiesrdquo Journal of the Franklin Institute vol350 no 6 pp 1562ndash1578 2013
[46] Y Ding H Zhu S Zhong and Y Zhang ldquo1198712minus 119871infin
filteringfor Markovian jump systems with time-varying delays andpartly unknown transition probabilitiesrdquo Communications inNonlinear Science and Numerical Simulation vol 17 no 7 pp3070ndash3081 2012
[47] Q Ma S Xu and Y Zou ldquoStability and synchronizationfor Markovian jump neural networks with partly unknowntransition probabilitiesrdquo Neurocomputing vol 74 pp 3403ndash3411 2011
[48] E Tian D Yue and G Wei ldquoRobust control for Markovianjump systems with partially known transition probabilities andnonlinearitiesrdquo Journal of the Franklin Institute vol 350 no 8pp 2069ndash2083 2013
[49] Y Guo and Z Wang ldquoStability of Markovian jump systemswith generally uncertain transition ratesrdquo Journal of the FranklinInstitute vol 350 no 9 pp 2826ndash2836 2013
[50] L Zhang ldquo119867infinestimation for discrete-time piecewise homoge-
neous Markov jump linear systemsrdquo Automatica vol 45 no 11pp 2570ndash2576 2009
[51] Z Wu H Su and J Chu ldquostate estimation for discreteMarkovian jumping neural networks with time delayrdquo Neuro-computing vol 73 pp 2247ndash2254 2010
[52] I R Petersen and A V Savkin Robust Kalman Filteringfor Signals and Systems with Large Uncertainties BirkhauserBoston Mass USA 1999
[53] B D O Anderson and J B Moore Optimal Filtering Prentice-Hall Upper Saddle River NJ USA 1979
[54] I R Petersen and D C McFarlane ldquoOptimal guaranteedcost control and filtering for uncertain linear systemsrdquo IEEETransactions on Automatic Control vol 39 no 9 pp 1971ndash19771994
[55] S Xu T Chen and J Lam ldquoRobust 119867infin filtering for uncertainMarkovian jump systems with mode-dependent time delaysrdquoIEEE Transactions on Automatic Control vol 48 no 5 pp 900ndash907 2003
[56] J H Kim D Oh and H Park ldquoGuaranteed cost and 119867infin
filtering for time-delay systemsrdquo in Proceeding of the AmericanControl Conference pp 4014ndash4018 IEEE Arlington Va USA2001
[57] L Xie ldquoOutput feedback119867infincontrol of systems with parameter
uncertaintyrdquo International Journal of Control vol 63 no 4 pp741ndash750 1996
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
2 Abstract and Applied Analysis
delay and partially unknown transition probabilities Guoand Wang [49] proposed another description for the uncer-tain TRs which is called generally uncertain TRs (GUTRs)
On the other hand state estimation plays an importantrole in systems and control theory signal processing andinformation fusion [50 51] Certainly the most widely usedestimation method is the well-known Kalman filtering [5253] A common feature in the Kalman filtering is that anaccurate model is available In some applications howeverwhen the system is subject to parameter uncertainties theaccurate system model is hard to obtain To overcome thisdifficulty the guaranteed cost filtering approach has beenproposed to ensure the upper bound of guaranteed cost func-tion [54] Robust119867
infinfiltering for uncertainMarkovian jump
systems with mode-dependent time delays was proposed in[55] In [56] guaranteed cost and 119867
infinfiltering for time-
delay systems were presented in terms of LMIs Howeverto the best of our knowledge there are few consideringthe robust guaranteed cost observer for a class of linearsingular Markovian jump time-delay systems with generallyincomplete transition probability which is still an openproblem
In this paper based on LMI method we address thedesign problem of the robust guaranteed cost observer for aclass of uncertain descriptor time-delay systems withMarko-vian jumping parameters and generally uncertain transitionrates The design problem proposed here is to design a mem-oryless observer such that for all uncertainties includinggenerally uncertain transition rates the resulting augmentedsystem is regular impulse-free and robust stochastically sta-ble and satisfies the proposed guaranteed cost performance
2 Problem Formulation
Consider the following descriptor time-delay systems withMarkovian jumping parameters
119864 (119905) = 119860 (119903119905 119905) 119909 (119905) + 119860
119889(119903119905 119905) 119909 (119905 minus 119889)
119910 (119905) = 119862 (119903119905 119905) 119909 (119905) + 119862
119889(119903119905 119905) 119909 (119905 minus 119889)
119909 (119905) = 120593 (119905) forall119905 isin [minus119889 0]
(1)
where 119909(119905) isin 119877119899 and 119910(119905) isin 119877
119903 are the state vector andthe controlled output respectively 119889 represents the state timedelay For convenience the input terms in system (1) havebeen omitted 120593(119905) isin 119871
2[minus119889 0] is a continuous vector-valued
initial function The random parameter 120574(119905) represents acontinuous-time discrete-state Markov process taking valuesin a finite set S = 1 2 119904 and having the transitionprobability matrix Π = [120587
119894119895] 119894 119895 isin 119873 The transition
probability from mode 119894 to mode 119895 is defined by
Pr 119903119905+Δ
= 119895 | 119903119905= 119894 =
120587119894119895Δ + 119900 (Δ) 119894 = 119895
1 + 120587119894119895Δ + 119900 (Δ) 119894 = 119895
(2)
where Δ gt 0 satisfies limΔrarr0
(119900(Δ)Δ) = 0 120587119894119895ge 0 is the
transition probability from mode 119894 to mode 119895 and satisfies
120587119894119894= minus
119904
sum
119895=1119895 = 119894
120587119894119895le 0 (3)
In this paper the transition rates of the jumping processare assumed to be partly available that is some elements inmatrix Λ have been exactly known some have been merelyknown with lower and upper bounds and others may haveno information to use For instance for system (1) withfour operation modes the transition rate matrix might bedescribed by
Λ =
[[[[
[
11+ Δ11
sdot sdot sdot
23+ Δ23
sdot sdot sdot 2119904+ Δ2119904
d
1199042+ Δ1199042
sdot sdot sdot
]]]]
]
(4)
where 119894119895and Δ
119894119895isin [minus120590
119894119895 120590119894119895] (120590119894119895
ge 0) represent theestimate value and estimate error of the uncertain TR 120587
119894119895
respectively where 119894119895and 120590
119894119895are known represents the
complete unknown TR which means that its estimate value119894119895and estimate error bound are unknownFor notational clarity for all 119894 isin S the set 119880
119894
denotes 119880119894
= 119880119894
119896cup 119880119894
119906119896with 119880
119894
119896= 119895 The estimate
value of 120587119894119895is known for 119895 isin S 119880119894
119906119896= 119895 The estimate
value of 120587119894119895is unknown for 119895 isin S Moreover if 119880119894
119896= 0 it is
further described as 119880119894119896= 119896119894
1 119896119894
2 119896
119894
119898 where 119896119894
119898isin N+
represents the 119898th bound-known element with the index119896119894
119898in the 119894th row of matrix Π We assume that the known
estimate values of the TRs are well defined That is
Assumption 1 If 119880119894119896
= S then 119894119895minus 120590119894119895
ge 0 (for all 119895 isin
S 119895 = 119894) 119894119894= minussum
119873
119895=1119895 = 119894119894119895and 120590
119894119894= minussum
119904
119895=1119895 = 119894120590119894119895
Assumption 2 If 119880119894
119896=S and 119894 isin 119880
119894
119896 then
119894119895minus 120590119894119895
ge
0 (for all 119895 isin S 119895 = 119894) 119894119894+ 120590119894119894le 0 and sum
119895isin119880119894
119896
119894119895
Assumption 3 If 119880119894
119896=S and 119894 notin 119880
119894
119896 then
119894119895minus 120590119894119895
ge
0 (for all 119895 isin S)
Remark 4 The above assumption is reasonable since it isthe direct result from the properties of the TRs (eg 120587
119894119895ge
0 (for all 119894 119895 isin S 119895 = 119894) and 120587119894119894= minussum
119904
119895=1119895 = 119894120587119894119895) The above
description about uncertain TRs is more general than eitherthe MJSs model with bounded uncertain TRs or the MJSsmodel with partly uncertain TRs If 119880119894
119906119896= 0 for all 119894 isin S
then generally uncertain TR matrix (4) reduces to boundeduncertain TR matrix (5) as follows
[[[[
[
11+ Δ11
12+ Δ12
sdot sdot sdot 1119904+ Δ1119904
21+ Δ21
22+ Δ22
sdot sdot sdot 2119904+ Δ2119904
d
1199041+ Δ1199041
1199042+ Δ1199042
sdot sdot sdot 119904119904+ Δ119904119904
]]]]
]
(5)
where 119894119895minusΔ119894119895ge 0 (for all 119895 isin S 119895 = 119894)
119894119894= minussum
119904
119895=1119895 = 119894119894119895le
0 and Δ119894119894= sum119904
119895=1119895 = 119894Δ119894119895 if 120590119894119895= 0 for all 119894 isin S for all 119895 isin
Abstract and Applied Analysis 3
119880119894
119896 then generally uncertain TR matrix (4) reduces to partly
uncertain TR matrix (6) as follows
[[[[
[
12058711
sdot sdot sdot
12058723
sdot sdot sdot 1205872119904
d
1205871199042
sdot sdot sdot
]]]]
]
(6)
Our results in this paper can be applicable to the generalMarkovian jump systems with bounded uncertain or partlyuncertain TR matrix
119860(120574(119905) 119905)119860119889(120574(119905) 119905)119862(120574(119905) 119905) and119862
119889(120574(119905) 119905) arematrix
functions of the random jumping process 120574(119905) To simplify thenotion the notation 119860
119894(119905) represents 119860(120574(119905) 119905) when 120574(119905) =
119894 For example 119860119889(120574(119905) 119905) is denoted by 119860
119889119894(119905) and so on
Further for each 120574(119905) = 119894 isin 119873 it is assumed that thematrices 119860
119894(119905) 119860
119889119894(119905) 119862119894(119905) and 119862
119889119894(119905) can be described by
the following form
119860119894(119905) = 119860
119894+ Δ119860119894(119905) 119860
119889119894(119905) = 119860
119889119894+ Δ119860119889119894(119905)
119862119894(119905) = 119862
119894+ Δ119862119894(119905) 119862
119889119894(119905) = 119862
119889119894+ Δ119862119889119894(119905)
(7)
where 119860119894 119860119889119894 119862119894are 119862
119889119894known real coefficient matrices
with appropriate dimensions Time-varying matricesΔ119860119894(119905) Δ119860
119889119894(119905) Δ119862
119894(119905) and Δ119862
119889119894(119905) represent norm-
bounded uncertainties and satisfy
[Δ119860119894(119905) Δ119860
119889119894(119905)
Δ119862119894(119905) Δ119862
119889119894(119905)
] = [1198721119894
1198722119894
]119865119894(119905) [1198731119894 1198732119894] (8)
where119872111989411987221198941198721119894 and119873
2119894are known constant real matri-
ces of appropriate dimensions which represent the structureof uncertainties and 119865
119894(119905) is an unknown matrix function
with Lebesgue measurable elements and satisfies 119865119894(119905)119865119879
119894(119905) le
119868Further for convenience we assume that the system has
the same dimension at each mode and the Markov processis irreducible Consider the following nominal unforceddescriptor time-delay system
119864 (119905) = 119860119894119909 (119905) + 119860
119889119894119909 (119905 minus 119889)
119909 (119905) = 120593 (119905) forall119905 isin [minus119889 0]
(9)
Let 1199090 1199030 and 119909(119905 120593 119903
0) be the initial state initial mode
and the corresponding solution of the system (9) at time 119905respectively
Definition 5 System (9) is said to be stochastically stable iffor all 120593(119905) isin 119871
2[minus119889 0] and initial mode 119903
0isin 119873 there exists
a matrix119872 gt 0 such that
119864int
infin
0
1003817100381710038171003817119909 (119905 120593 1199030)10038171003817100381710038172
119889119905 | 1199030 119909 (119905) = 120593 (119905) 119905 isin [minus119889 0]
le 119909119879
01198721199090
(10)
The following definition can be regarded as an extensionof the definition in [2]
Definition 6 (1) System (9) is said to be regular if det(119904E minus
119860119894) 119894 = 1 2 119904 are not identically zero(2) System (9) is said to be impulse free if deg(det(119904119864 minus
119860119894)) = rank 119864
119894 119894 = 1 2 119904
(3) System (9) is said to be admissible if it is regularimpulse free and stochastically stable
The linearmemoryless observer under consideration is asfollows
119864 119909 (119905) = 1198701119894119909 (119905) + 119870
2119894119910 (119905)
1199090= 0 119903 (0) = 119903
0
(11)
where 119909(119905) isin 119877119899 is the observer state and the constant
matrices1198701119894and119870
2119894are observer parameters to be designed
Denote the error state 119890(119905) = 119909(119905) minus 119909(119905) and theaugmented state vector 119909
119891= [119909119879(119905) 119890119879(119905)]119879
Let 119909(119905) = 119871119890(119905)
represent the output of the error states where 119871 is a knownconstant matrix Define
119860119891119894= [
119860119894
0
119860119894minus 1198701119894minus 11987021198941198621198941198701119894
]
119860119891119889119894
= [119860119889119894
0
119860119889119894minus 1198702119894119862119889119894
0] 119864
119891= [
119864 0
0 119864]
119872119891119894= 1198721198911119894
= [1198721119894
1198721119894minus 11987021198941198722119894
] 119873119891119894= [1198731119894 0]
Δ119860119891119894= 119872119891119894119865119894(119905)119873119891119894 119873
1198911119894= [1198732119894 0]
Δ119860119891119889119894
= 1198721198911119894
119865119894(119905)1198731198911119894
119862119891= [0 119871]
(12)
and combine (1) and (11) then we derive the augmentedsystems as follows
119864119891119891(119905) = (119860
119891119894+ Δ119860119891119894) 119909119891(119905)
+ (119860119891119889119894
+ Δ119860119891119889119894
) 119909119891(119905 minus 119889)
119911 (119905) = 119862119891119909119891(119905)
1199091198910
(119905) = [120593119879(119905) 120593119879(119905)]119879
forall119905 isin [minus119889 0]
(13)
Similar to [5] it is also assumed in this paper that for all 120589 isin[minus119889 0] there exists a scalar ℎ gt 0 such that 119909
119891(119905 + 120589) le
ℎ119909119891(119905)Associated with system (13) is the cost function
J = Eint
infin
0
119911119879(119905) 119911 (119905) 119889119905 (14)
Definition 7 Consider the augmented system (13) if thereexist the observer parameters 119870
1119894 1198702119894and a positive scalar
Jlowast for all uncertainties such that the augmented system(13) is robust stochastically stable and the value of the costfunction (14) satisfies J le Jlowast then Jlowast is said to be arobust guaranteed cost and observer (11) is said to be a robustguaranteed cost observer for system (1) with (4)
Problem 8 (robust guaranteed cost observer problem for aclass of linear singular Markovian jump time-delay systems
4 Abstract and Applied Analysis
with generally incomplete transition probability) Given sys-tem (1) with GUTRMatrix (4) can we determine an observer(11) with parameters 119870
1119894and 119870
2119894such that the observer is a
robust guaranteed cost observer for system (1) with GUTRMatrix (4)
Lemma 9 Given any real number 120576 and any matrix Q thematrix inequality 120576(119876 + 119876
119879) le 1205762119879 + 119876119879
minus1119876119879 holds for any
matrix 119879 gt 0
3 Main Results
Theorem 10 Consider the augmented system (13)with GUTRMatrix (4) and the cost function (14) Then the robust guar-anteed cost observer (11) with parameters 119870
1119894and 119870
2119894can
be designed if there exist matrices 119875119894 1198701119894 and 119870
2119894 119894 =
1 2 119904 and symmetric positive definite matrix Q satisfyingthe following LMIs respectively
Case 1 If 119894 notin 119880119894
119896and 119880
119894
119896= 119896119894
1 119896
119894
119898 there exist a set of
symmetric positive definite matrices 119879119894119895isin R119899times119899 (119894 notin 119880
119894
119896 119895 isin
119880119894
119896) such that
119864119879
119891119875119894= 119875119879
119894119864119891ge 0 (15)
[[
[
Π119894+ 119862119879
119891119862119891
119875119894(119860119891119889119894
+ Δ119860119891119889119894
) 1
(119860119891119889119894
+ Δ119860119891119889119894
)119879
119875119894
minus119876 0
lowast lowast 2
]]
]
lt 0 (16)
119875119894minus 119875119895ge 0 forall119895 isin 119880
119894
119906119896 119895 = 119894 (17)
Case 2 If 119894 isin 119880119894
119896 119880119894119896= 119896119894
1 119896
119894
119898 and 119880119894
119906119896= 0 there exist a
set of symmetric positive definite matrices 119881119894119895119897
isin R119899times119899 (119894 119895 isin
119880119894
119896 119897 isin 119880
119894
119906119896) such that
119864119879
119891119875119894= 119875119879
119894119864119891ge 0 (18)
[[
[
Ω119894+ 119862119879
119891119862119891
119875119894(119860119891119889119894
+ Δ119860119891119889119894
) 1
(119860119891119889119894
+ Δ119860119891119889119894
)119879
119875119894
minus119876 0
lowast lowast 2
]]
]
lt 0 (19)
Case 3 If 119894 isin 119880119894
119896and 119880
119894
119906119896= 0 there exist a set of symmetric
positive definite matrices119882119894119895isin R119899times119899 (119894 119895 isin 119880
119894
119896) such that
119864119879
119891119875119894= 119875119879
119894119864119891ge 0 (20)
[[
[
Δ119894+ 119862119879
119891119862119891
119875119894(119860119891119889119894
+ Δ119860119891119889119894
) 1
(119860119891119889119894
+ Δ119860119891119889119894
)119879
119875119894
minus119876 0
lowast lowast 2
]]
]
lt 0 (21)
where
Π119894= (119860119891119894+ Δ119860119891119894)119879
119875119894+ 119875119894(119860119891119894+ Δ119860119891119894)
+ 119876 + sum
119895isin119880119894
119896
119894119895119864119879
119891(119875119895minus 119875119894) + sum
119895isin119880119894
119896
1
41205902
119894119895119879119894119895
Ω119894= (119860119891119894+ Δ119860119891119894)119879
119875119894+ 119875119894(119860119891119894+ Δ119860119891119894)
+ 119876 + sum
119895isin119880119894
119896
119894119895119864119879
119891(119875119895minus 119875119897) + sum
119895isin119880119894
119896
1
41205902
119894119895119881119894119895119897
Δ119894= (119860119891119894+ Δ119860119891119894)119879
119875119894+ 119875119894(119860119891119894+ Δ119860119891119894)
+ 119876 + sum
119895isinS119895 = 119894
119894119895119864119879
119891(119875119895minus 119875119894) + sum
119895isinS119895 = 119894
1
41205902
119894119895119882119894119895
1= [119864119879
119891(119875119894119896119894
1
minus 119875119894) 119864119879
119891(119875119894119896119894
2
minus 119875119894) 119864
119879
119891(119875119894119896119894
119898
minus 119875119894)]
2= diag minus119879
119894119896119894
1
minus119879119894119896119894
119898
1= [119864119879
119891(119875119896119894
1
minus 119875119897) 119864119879
119891(119875119896119894
2
minus 119875119897) 119864
119879
119891(119875119896119894
119898
minus 119875119897)]
2= diag minus119881
119894119896119894
1119897 minus119881
119894119896119894
119898119897
1= [119864119879
119891(1198751minus 119875119894) 119864
119879
119891(119875119894minus1
minus 119875119894)
119864119879
119891(119875119894+1
minus 119875119894) 119864
119879
119891(119875119904minus 119875119894)]
2= diag minus119882
1198941 minus119882
119894119904
(22)
Proof According to Definition 2 and Theorem 1 in [2] wecan derive from (15)ndash(21) that system (13) is regular andimpulse free Let the mode at time 119905 be 119894 and consider thefollowing Lyapunov function with respect to the augmentedsystem (13)
119881(119909119891(119905) 120574 (119905) = 119894) = 119909
119879
119891(119905) 119864119879
119891119875119894119909119891(119905)
+ int
119905
119905minus119889
119909119879
119891(119904) 119876119909
119891(119904) 119889119905
(23)
where 119876 is the symmetric positive definite matrix to be cho-sen and 119875
119894is a matrix satisfying (15)ndash(21)The weak infinites-
imal operatorL of the stochastic process 120574(119905) 119909119891(119905) 119905 ge 0
is presented by
L119881(119909119891(119905) 120574 (119905) = 119894)
= limΔrarr0
1
Δ[119864119891119881 (119909 (119905 + Δ) 120574 (119905 + Δ)) 119909 (119905) 120574 (119905) = 119894
minus119881 (119909 (119905) 120574 (119905) = 119894) ]
= 119909119879
119891(119905) [
[
(119860119891119894+ Δ119860119891119894)119879
119875119894+ 119875119894(119860119891119894+ Δ119860119891119894)
+
119904
sum
119895=1
120587119894119895119864119879
119891119875119895+ 119876]
]
119909119891(119905)
+ 2119909119879
119891(119905) 119875119894(1198601198911119894
+ Δ1198601198911119894
) 119909119891(119905 minus 119889)
minus 119909119879
119891(119905 minus 119889)119876119909
119891(119905 minus 119889)
(24)
Abstract and Applied Analysis 5
Case 1 (119894 notin 119880119894
119896) Note that in this case sum
119895isin119880119894
119906119896119895 = 119894
120587119894119895
=
minussum119895isinU119894119896119895 = 119894
120587119894119895minus 120587119894119894and 120587
119894119895ge 0 119895 isin 119880
119894
119906119896 119895 = 119894 then from
(24) we have
119909119879
119891(119905) [
[
119904
sum
119895=1
120587119894119895119864119879
119891119875119895]
]
119909119891(119905)
= 119909119879
119891(119905) [
[
sum
119895isin119880119894
119896
120587119894119895119864119879
119891119875119895+ sum
119895isin119880119894
119906119896119895 = 119894
120587119894119895119864119879
119891119875119895+ 120587119894119894119864119879
119891119875119895]
]
119909119891(119905)
= 119909119879
119891(119905) [
[
sum
119895isin119880119894
119896
120587119894119895119864119879
119891119875119895+ (minus120587
119894119894minus sum
119895isin119880119894
119896
120587119894119895)119864119879
119891119875119894
+120587119894119894119864119879
119891119875119894]
]
119909119891(119905)
= 119909119879
119891(119905) 119864119879
119891[
[
sum
119895isin119880119894
119896
120587119894119895(119875119895minus 119875119894)]
]
119909119891(119905)
= 119909119879
119891(119905) 119864119879
119891[
[
sum
119895isin119880119894
119896
119894119895(119875119895minus 119875119894) + sum
119895isin119880119894
119896
Δ119894119895(119875119895minus 119875119894)]
]
119909119891(119905)
(25)
On the other hand in view of Lemma 9 we have
sum
119895isin119880119894
119896
Δ119894119895119864119879
119891(119875119895minus 119875119894)
= sum
119895isin119880119894
119896
[1
2Δ119894119895119864119879
119891(119875119895minus 119875119894) +
1
2Δ119894119895119864119879
119891(119875119895minus 119875119894)]
le sum
119895isin119880119894
119896
[(1
2Δ119894119895)
2
119879119894119895+ 119864119879
119891(119875119895minus 119875119894) 119879minus1
119894119895(119875119895minus 119875119894) 119864119891]
le sum
119895isin119880119894
119896
[1
41205902
119894119895119879119894119895+ 119864119879
119891(119875119895minus 119875119894) 119879minus1
119894119895(119875119895minus 119875119894) 119864119891]
(26)
Case 2 (119894 isin 119880119894
119896and119880119894
119906119896= 0) Because of119880119894
119896= 119896119894
1 119896
119894
119898 and
119880119894
119906119896= 119906119894
1 119906
119894
119904minus119898 there must be 119897 isin 119880
119894
119906119896so that 119864119879
119891119875119897ge
119864119879
119891119875119895(for all 119895 isin 119880
119894
119906119896)
119909119879
119891(119905) [
[
119904
sum
119895=1
120587119894119895119864119879
119891119875119895]
]
119909119891(119905)
le 119909119879
119891(119905) [
[
sum
119895isin119880119894
119896
120587119894119895119864119879
119891119875119895minus ( sum
119895isin119880119894
119896119895 = 119894
120587119894119895)119864119879
119891119875119897]
]
119909119891(119905)
= 119909119879
119891(119905) 119864119879
119891[
[
sum
119895isin119880119894
119896
120587119894119895(119875119895minus 119875119897)]
]
119909119891(119905)
= 119909119879
119891(119905) 119864119879
119891[
[
sum
119895isin119880119894
119896
119894119895(119875119895minus 119875119897) + sum
119895isin119880119894
119896
Δ119894119895(119875119895minus 119875119897)]
]
119909119891(119905)
(27)
By using Lemma 9 we have
sum
119895isin119880119894
119896
Δ119894119895119864119879
119891(119875119895minus 119875119897)
= sum
119895isin119880119894
119896
[1
2Δ119894119895119864119879
119891(119875119895minus 119875119897) +
1
2Δ119894119895119864119879
119891(119875119895minus 119875119897)]
le sum
119895isin119880119894
119896
[(1
2Δ119894119895)
2
119881119894119895119897+ 119864119879
119891(119875119895minus 119875119897)119881minus1
119894119895119897(119875119895minus 119875119897) 119864119879
119891]
le sum
119895isin119880119894
119896
[1
41205902
119894119895119881119894119895119897+ 119864119879
119891(119875119895minus 119875119897)119881minus1
119894119895119897(119875119895minus 119875119897) 119864119879
119891]
(28)
Case 3 (119894 isin 119880119894
119896and 119880
119894
119906119896= 0) Consider
119909119879
119891(119905) [
[
119904
sum
119895=1
120587119894119895119864119879
119891119875119895]
]
119909119891(119905)
= 119909119879
119891(119905) 119864119879
119891[
[
119904
sum
119895=1119895 = 119894
120587119894119895(119875119895minus 119875119894)]
]
119909119891(119905)
= 119909119879
119891(119905) 119864119879
119891[
[
119904
sum
119895=1119895 = 119894
119894119895(119875119895minus 119875119894)
+
119904
sum
119895=1119895 = 119894
Δ119894119895(119875119895minus 119875119894)]
]
119909119891(119905)
(29)
Case 1 Substituting (25) and (26) into (24) it results in
L119881 le Λ119879(119905) Φ (119894) Λ (119905) (30)
where Λ119879(119905) = [119909119879
119891(119905) 119909119879
119891(119905 minus 119889)] and
Φ119894=[[
[
Π119894+ 119862119879
119891119862119891
119875119894(119860119891119889119894
+ Δ119860119891119889119894
) 1
(119860119891119889119894
+ Δ119860119891119889119894
)119879
119875119894
minus119876 0
lowast lowast 2
]]
]
(31)
Case 2 Substituting (27) and (28) into (24) it results in
L119881 le Λ119879(119905) Ψ (119894) Λ (119905) (32)
where Λ119879(119905) = [119909119879
119891(119905) 119909119879
119891(119905 minus 119889)] and
Ψ119894=[[
[
Ω119894+ 119862119879
119891119862119891
119875119894(119860119891119889119894
+ Δ119860119891119889119894
) 1
(119860119891119889119894
+ Δ119860119891119889119894
)119879
119875119894
minus119876 0
lowast lowast 2
]]
]
(33)
6 Abstract and Applied Analysis
Case 3 Substituting (29) into (24) we get
L119881 le Λ119879(119905) Γ (119894) Λ (119905) (34)
where Λ119879(119905) = [119909119879
119891(119905) 119909119879
119891(119905 minus 119889)] and
Γ119894=[[
[
Δ119894+ 119862119879
119891119862119891
119875119894(119860119891119889119894
+ Δ119860119891119889119894
) 1
(119860119891119889119894
+ Δ119860119891119889119894
)119879
119875119894
minus119876 0
lowast lowast 2
]]
]
(35)
Similar to [5] usingDynkinrsquos formula we drive for each 119894 isin 119873
lim119879rarrinfin
Eint
119879
0
119909119879
119891(119905) 119909119891(119905) 119889119905 | 120593
119891 1205740= 119894 le 119909
119879
11989101198721199091198910 (36)
By Definition 5 it is easy to see that the augmented system(13) is stochastically stable Furthermore from (16) (19) and(21) we have
L119881 le minus119909119879
119891(119905) 119862119879
119891119862119891119909119891(119905) lt 0 (37)
On the other hand we have
J = Eint
infin
0
119909119879
119891(119905) 119862119879
119891119862119891119909119891(119905) 119889119905 lt minusint
infin
0
L119881119889119905
= minus E lim119905rarrinfin
119881 (119909 (119905) 120574 (119905)) + 119881 (1199090 1205740)
(38)
As the augmented system (13) is stochastically stable itfollows from (38) that 119869 lt 119881(119909
1198910 1199030) From Definition 7 it
is concluded that a robust guaranteed cost for the augmentedsystem (13) can be given by 119869
lowast= 119909119879
1198910(119905)119864119879
1198911199030119875(1199030)1199091198910
+
int0
minus119889119909119879
119891(119905)119876119909
119891(119905)119889119905
In the following based on the above sufficient conditionthe design of robust guaranteed cost observers can be turnedinto the solvability of a system of LMIs
Theorem 11 Consider system (13)with GUTRMatrix (4) andthe cost function (14) If there exist matrices 119884
1119894and 119884
2119894 119894 =
1 2 119904 positive scalars 120576119894 119894 = 1 2 119904 symmetric positive
definite matrix Q and the full rank matrices 1198752119894 and matrices
119875119894= diag(119875
1119894 1198752119894) 119894 = 1 2 119904 satisfying the following
LMIs respectively
Case 1 If 119894 notin 119880119894
119896and 119880
119894
119896= 119896119894
1 119896
119894
119898 a set of positive
definite matrices 119879119894119895isin R119899times119899 (119894 notin 119880
119894
119896 119895 isin 119880
119894
119896) exist such that
119864119879
119891119875119894= 119875119879
119894119864119891ge 0 (39)
[[[[[[[[[
[
1206011119894
1206012119894
1198731
1206013119894
120601119879
2119894minus119876 0 0
119873119879
10 119873
20
120601119879
31198940 0 minus120576
119894119868
]]]]]]]]]
]
lt 0 (40)
119875119894minus 119875119895ge 0 forall119895 isin 119880
119894
119906119896 119895 = 119894 (41)
Case 2 If 119894 isin 119880119894
119896(119880119894
119896= 119896119894
1 119896
119894
119898) and 119880
119894
119906119896= 0 a set of
positive definite matrices119881119894119895119897isin R119899times119899 (119894 119895 isin 119880
119894
119896 119897 isin 119880
119894
119906119896) exist
such that
119864119879
119891119875119894= 119875119879
119894119864119891ge 0 (42)
[[[[[[[[[
[
1205931119894
1205932119894
1198721
1205933119894
120593119879
2119894minus119876 0 0
119872119879
10 119872
20
120593119879
31198940 0 minus120576
119894119868
]]]]]]]]]
]
lt 0 (43)
Case 3 If 119894 isin 119880119894
119896and 119880
119894
119906119896= 0 a set of positive definite
matrices119882119894119895isin R119899times119899 (119894 119895 isin 119880
119894
119896) exist such that
119864119879
119891119894119875119894= 119875119879
119894119864119891119894ge 0 (44)
[[[[[[[[[
[
1205951119894
1205952119894
1198711
1205953119894
120595119879
2119894minus119876 0 0
119871119879
10 1198712
0
120595119879
31198940 0 minus120576
119894119868
]]]]]]]]]
]
lt 0 (45)
where
1206011119894= 1205931119894= 1205951119894
= [1198751119894119860119894+ 119860119879
1198941198751119894
119860119879
1198941198752119894minus 119884119879
1119894minus 119862119879
119894119884119879
2119894
1198752119894119860119894minus 1198841119894minus 1198842119894119862119894
119884119879
1119894+ 1198841119894
]
+ 119876 + 119862119879
119891119862119891+ sum
119895isin119880119894
119896
119894119895119864119879
119891(119875119895minus 119875119894) + sum
119895isin119880119894
119896
1
41205902
119894119895119879119894119895
1206012119894= 1205932119894= 1205952119894= [
1198751119894119860 i 0
1198752119894119860119894minus 1198841119894minus 11988421198941198621198940]
1206013119894= 1205933119894= 1205953119894= [
11987511198941198721119894
11987521198941198721119894minus 11988411198941198721119894minus 11988421198941198722119894
]
1198731= [119864119879
119891(119875119896119894
1
minus 119875119894) 119864119879
119891(119875119896119894
2
minus 119875119894) 119864
119879
119891(119875119896119894
119898
minus 119875119894)]
1198732= diag minus119879
119894119896119894
1
minus119879119894119896119894
119898
1198721= [119864119879
119891(119875119896119894
1
minus 119875119894) 119864119879
119891(119875119896119894
2
minus 119875119894) 119864
119879
119891(119875119896119894
119898
minus 119875119894)]
1198722= diag minus119881
119894119896119894
1119897 minus119881
119894119896119894
119898119897
1198711= [119864119879
119891(1198751minus 119875119894) 119864
119879
119891(119875119894minus1
minus 119875119894)
119864119879
119891(119875119894+1
minus 119875119894) 119864
119879
119891(119875119904minus 119875119894)]
1198712= diag minus119882
1198941 minus119882
119894119904
(46)
Abstract and Applied Analysis 7
Then a suitable robust guaranteed cost observer in theform of (11) has parameters as follows
1198701119894= 119875minus1
11198941198841119894 119870
2119894= 119875minus1
21198941198842119894
(47)
and 119869lowast is a robust guaranteed cost for system (13) with GUTRMatrix (4)
Proof Define
1198601
119894=
[[[[
[
119860119879
119891119894119875119894+ 119875119894119860119891119894+ 119876 + sum
119895isin119880119894
119896
119894119895(119875119895minus 119875119894) + sum
119895isin119880119894
119896
1
41205902
119894119895119879119894119895+ 119862119879
119891119862119891
119875119894119860119891119889119894
1198731
119860119879
119891119889119894119875119894
minus119876 0
lowast lowast 1198732
]]]]
]
(48)
1198602
119894=
[[[[
[
119860119879
119891119894119875119894+ 119875119894119860119891119894+ 119876 + sum
119895isin119880119894
119896
119894119895(119875119895minus 119875119894) + sum
119895isin119880119894
119896
1
41205902
119894119895119881119894119895119897+ 119862119879
119891119862119891
119875119894119860119891119889119894
1198721
119860119879
119891119889119894119875119894
minus119876 0
lowast lowast 1198722
]]]]
]
(49)
1198603
119894=
[[[[
[
119860119879
119891119894119875119894+ 119875119894119860119891119894+ 119876 + sum
119895isin119880119894
119896
119894119895(119875119895minus 119875119894) + sum
119895isin119880119894
119896
1
41205902
119894119895119882119894119895+ 119862119879
119891119862119891
119875119894119860119891119889119894
1198711
119860119879
119891119889119894119875119894
minus119876 0
lowast lowast 1198712
]]]]
]
lt 0 (50)
Then (16) is equivalent to
1198601
119894+ [
[
119875119894119872119891119894
0
0
]
]
119865119894[119873119891119894 1198731198911119894 0]
+ [119873119891119894 1198651198911119894 0]119879
119865119879
119894[
[
119875119894119872119891119894
0
0
]
]
119879
lt 0
(51)
Then (19) is equivalent to
1198602
119894+ [
[
119875119894119872119891119894
0
0
]
]
119865119894[119873119891119894 1198731198911119894 0]
+ [119873119891119894 1198651198911119894 0]119879
119865119879
119894[
[
119875119894119872119891119894
0
0
]
]
119879
lt 0
(52)
Then (21) is equivalent to
1198603
119894+ [
[
119875119894119872119891119894
0
0
]
]
119865119894[119873119891119894 1198731198911119894 0]
+ [119873119891119894 1198651198911119894 0]119879
119865119879
119894[
[
119875119894119872119891119894
0
0
]
]
119879
lt 0
(53)
By applying Lemma 24 in [57] (50) (51) and (52) hold forall uncertainties119865i satisfying119865
119879
119894119865119894lt 119868 if and only if there exist
positive scalars 120576119894 119894 = 1 2 119904 such that
1198601
119894+ 120576minus1
119894[
[
119875119894119872119891119894
0
0
]
]
[
[
119875119894119872119891119894
0
0
]
]
119879
+ 120576119894[119873119891119894 1198651198911119894 0]
119879
[119873119891119894 1198651198911119894 0] lt 0
1198602
119894+ 120576minus1
119894[
[
119875119894119872119891119894
0
0
]
]
[
[
119875119894119872119891119894
0
0
]
]
119879
+ 120576119894[119873119891119894 1198651198911119894 0]
119879
[119873119891119894 1198651198911119894 0] lt 0
1198603
119894+ 120576minus1
119894[
[
119875119894119872119891119894
0
0
]
]
[
[
119875119894119872119891119894
0
0
]
]
119879
+ 120576119894[119873119891119894 1198651198911119894 0]
119879
[119873119891119894 1198651198911119894 0] lt 0
(54)
Let 119875119894= diag(119875
1119894 1198752119894) and using (47) we can conclude from
Schur complement results that the above matrix inequalitiesare equivalent to the coupled LMIs (40) (43) and (45)It further follows from Theorem 10 that 119869
lowast is a robustguaranteed cost for system (13) with (4)
Remark 12 The solution of LMIs (39)ndash(45) parameterizesthe set of the proposed robust guaranteed cost observersThis parameterized representation can be used to design theguaranteed cost observer with some additional performanceconstraints By applying the methods in [14] the suboptimal
8 Abstract and Applied Analysis
guaranteed cost observer can be determined by solving acertain optimization problem This is the following theorem
Theorem 13 Consider system (13)with GUTRMatrix (4) andthe cost function (14) and suppose that the initial conditions 119903
0
and 1199091198910
are known if the following optimization problem
min1198761198751119894 1198752119894 120576119894 1198841119894 and 1198842119894
119869lowast
st LMIs (39) ndash (45)(55)
has a solution 119876 1198751119894 1198752119894 120576119894 1198841119894 and 119884
2119894 119894 = 1 2 119904
then the observer (11) is a suboptimal guaranteed costobserver for system (1) where 119869
lowast= 119909119879
1198910119864119879
1198911199030119875(1199030)1199091198910
+
tr(int0minus119889
1199091198910(119905)1199091198910(119905)119909119879
1198910119889119905119876)
Proof It follows from Theorem 11 that the observer (11)constructed in terms of the solution 119876 119875
1119894 1198752119894 120576119894 1198841119894 and
1198842119894 119894 = 1 2 119904 is a robust guaranteed cost observer By
noting that
int
0
minus119889
119909119879
1198910(119905) 119876119909
1198910(119905) 119889119905 = int
0
minus119889
tr (1199091198791198910
(119905) 1198761199091198910
(119905)) 119889119905
= tr(int0
minus119889
119909119879
1198910(119905) 1199091198910
(119905) 119889119905119876)
(56)
it follows that the suboptimal guaranteed cost observerproblem is turned into the minimization problem (55)
Remark 14 Theorem 13 gives the suboptimal guaranteed costobserver conditions of a class of linear Markovian jump-ing time-delay systems with generally incomplete transitionprobability and LMI constraints which can be easily solvedby the LMI toolbox in MATLAB
4 Numerical Example
In this section a numerical example is presented todemonstrate the effectiveness of the method mentioned inTheorem 11 Consider a 2-dimensional system (1) with 3Markovian switching modes In this numerical example thesingular system matrix is set as 119864 = [ 1 0
0 0] and the 3-mode
transition rate matrix is Λ = [minus32
2
15 21 minus36
] where Δ11 Δ31
isin
[minus015 015] Δ23 Δ33
isin [minus012 012] and Δ32
isin [minus01 01]The other system matrices are as follows
For mode 119894 = 1 there are
1198601= [
minus32 065
1 02] 119860
1198891= [
02 05
1 minus068]
1198621= [
[
12 065
minus65 19
minus021 minus18
]
]
1198621198891
= [
[
minus36 minus105
21 096
021 minus086
]
]
11987211
= [minus02
08] 119872
21= [
[
025
0875
minus2
]
]
11987311
= [minus12 31] 11987321
= [minus069 minus42]
(57)
For mode 119894 = 2 there are
1198602= [
minus1 6
2 minus36] 119860
1198892= [
minus31 minus16
3 075]
1198622= [
[
9 minus25
035 minus2
36 minus18
]
]
1198621198892
= [
[
089 minus6
minus12 09
minus24 6
]
]
11987212
= [23
minus4] 119872
22= [
[
075
minus36
25
]
]
11987312
= [minus72 minus6] 11987322
= [1 2]
(58)
For mode 119894 = 3 there are
1198603= [
minus106 29
minus03 36] 119860
1198893= [
minus56 minus12
minus3 45]
1198623= [
[
minus3 minus036
015 minus18
09 minus5
]
]
1198621198893
= [
[
minus165 5
minus12 265
minus098 minus56
]
]
11987213
= [minus82
minus03] 119872
23= [
[
minus052
25
minus36
]
]
11987313
= [105 minus5] 11987323
= [minus72 minus126]
(59)
Then we set the error state matrix 119871 = [ minus45 062 minus6
] and thepositive scalars in Theorem 11 are 120576
1= 02 120576
2= 015 120576
3=
032 According to the definitions of augmented statematricesin (12) we can easily obtain the following parameter matricesin Theorem 11 by MATLAB
11988411
= [minus84521006 00127
00127 84509001]
11988421
= [002 01291 01435
minus23520 minus04080 minus08106]
11988412
= [minus170991 269626
269626 minus246750]
11988422
= [minus200744 minus132941 529388
216893 180120 minus506693]
11988413
= [minus6751329 224456
224456 minus8976976]
11988423
= [minus136021 minus1461726 540500
minus34324 minus195125 minus103068]
1198751=[[[
[
63029 0 0 0
0 48620 0 0
0 0 22914 0
0 0 0 03169
]]]
]
Abstract and Applied Analysis 9
1198752=[[[
[
08914 0 0 0
0 12505 0 0
0 0 73629 0
0 0 0 30056
]]]
]
1198753=[[[
[
30265 0 0 0
0 02156 0 0
0 0 08965 0
0 0 0 10002
]]]
]
119876 =[[[
[
05000 0 0 0
0 05001 0 0
0 0 05001 0
0 0 0 05001
]]]
]
11987911
=[[[
[
34173214 minus8707765 0 0
minus8707765 4167216 0 0
0 0 22263598 minus3207456
0 0 minus3207456 12263101
]]]
]
11987923
=[[[
[
37753231 minus27999330 0 0
minus27999330 28107685 0 0
0 0 106907366 minus107432750
0 0 minus107432750 108555053
]]]
]
11987931
=[[[
[
9518504 minus5399245 0 0
minus5399245 8962029 0 0
0 0 14773012 minus2077540
0 0 minus2077540 14791256
]]]
]
11987932
=[[[
[
21617695 minus12094164 0 0
minus12094164 20371205 0 0
0 0 14786 minus09283
0 0 minus09283 14794
]]]
]
11987933
=[[[
[
14939313 minus8398780 0 0
minus8398780 14073689 0 0
0 0 1478123 minus1336452
0 0 minus1336452 2459347
]]]
]
11988111
=[[[
[
16650 0 0 0
0 16650 0 0
0 0 16650 0
0 0 0 16650
]]]
]
11988231
=[[[
[
15426 0 0 0
0 16650 0 0
0 0 16662 0
0 0 0 16650
]]]
]
11988232
=[[[
[
15428 0 0 0
0 16650 0 0
0 0 16622 0
0 0 0 16650
]]]
]
(60)
Therefore we can design a linear memoryless observer as(11) with the constant matrices
11987011
= 119875minus1
1111988411
= [minus13409860 00020
00026 17381530]
11987021
= 119875minus1
2111988421
= [00087 00563 00626
minus74219 minus12875 minus25579]
11987012
= 119875minus1
1211988412
= [minus191823 302475
215615 minus197321]
11987022
= 119875minus1
2211988422
= [minus27264 minus18056 71899
72163 59928 minus168583]
11987013
= 119875minus1
1311988413
= [minus2231402 74186
1041076 minus41637180]
11987023
= 119875minus1
2311988423
= [minus151724 minus1630481 602900
minus34317 minus195086 minus103047]
(61)
Finally the observer (11) with the above parameter matri-ces for this numerical example is a suboptimal guaranteedcost observer byTheorems 11 and 13
5 Conclusions
In this paper the robust guaranteed cost observer problemfor a class of uncertain descriptor time-delay systems withMarkovian jumping parameters and generally uncertaintransition rates is studied by using LMI method In thisGUTR singular model each transition rate can be com-pletely unknown or only its estimate value is known Theparameterrsquos uncertainty is time varying and is assumed tobe norm-bounded Memoryless guaranteed cost observersare designed in terms of a set of linear coupled matrixinequalities The suboptimal guaranteed cost observer isdesigned by solving a certain optimization problem Ourresults can be applicable to the general Markovian jumpsystems with bounded uncertain or partly uncertain TRmatrix
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This research is supported by the National NaturalScience Foundations of China (10701020 1097102260973048 61272077 60974025 60673101 and 60939003)NCET-08-0755 National 863 Plan Project (2008AA04Z401 2009AA043404) SDPW IMZQWH010016the Natural Science Foundation of Shandong Province (noZR2012FM006) the Scientific and Technological Projectof Shandong Province (no 2007GG3WZ04016) and theNatural Science Foundation of Guangxi Autonomous Region(no 2012GXNSFBA053003)
10 Abstract and Applied Analysis
References
[1] L Dai Singular Control Systems Springer Berlin Germany1989
[2] S Xu P V Dooren R Stefan and J Lam ldquoRobust stability andstabilization for singular systemswith state delay and parameteruncertaintyrdquo IEEE Transactions on Automatic Control vol 47no 7 pp 1122ndash1128 2002
[3] M Fu and G Duan ldquoRobust guaranteed cost observer foruncertain descriptor time-delay systems withMarkovian jump-ing parametersrdquo Acta Automatica Sinica vol 31 pp 479ndash4832005
[4] Y Q Xia E-K Boukas P Shi and J H Zhang ldquoStabilityand stabilization of continuous-time singular hybrid systemsrdquoAutomatica vol 45 no 6 pp 1504ndash1509 2009
[5] S Xu and J Lam Robust Control and Filtering of SingularSystems Springer Berlin Germany 2006
[6] Y-Y Cao and J Lam ldquoRobust 119867infin
control of uncertainMarkovian jump systems with time-delayrdquo IEEE Transactionson Automatic Control vol 45 no 1 pp 77ndash83 2000
[7] Y Li and Y Kao ldquoStability of stochastic reaction-diffusion sys-tems with Markovian switching and impulsive perturbationsrdquoMathematical Problems in Engineering vol 2012 Article ID429568 13 pages 2012
[8] X Mao and C Yuan Stochastic Differential Equations withMarkovian Switching Imperial College Press London UK2006
[9] E-K Boukas Stochastic Switching Systems Analysis and DesignBirkhauser Basel Switzerland 2005
[10] Y Kao C Wang and L Zhang ldquoDelay-dependent exponentialstability of impulsive markovian jumping Cohen-Grossbergneural networks with reaction-diffusion and mixed delaysrdquoNeural Processing Letters vol 38 no 3 pp 321ndash346 2013
[11] Y-G Kao J-F Guo C-H Wang and X-Q Sun ldquoDelay-dependent robust exponential stability of Markovian jump-ing reaction-diffusion Cohen-Grossberg neural networks withmixed delaysrdquo Journal of the Franklin Institute vol 349 no 6pp 1972ndash1988 2012
[12] Y Kao C Wang F Zha and H Cao ldquoStability in mean ofpartial variables for stochastic reactionmdashdiffusion systems withMarkovian switchingrdquo Journal of the Franklin Institute vol 351no 1 pp 500ndash512 2014
[13] N Kumaresan and P Balasubramaniam ldquoOptimal control forstochastic nonlinear singular system using neural networksrdquoComputers amp Mathematics with Applications vol 56 no 9 pp2145ndash2154 2008
[14] L Yu and J Chu ldquoAn LMI approach to guaranteed cost controlof linear uncertain time-delay systemsrdquoAutomatica vol 35 no6 pp 1155ndash1159 1999
[15] S Yin H Luo and S Ding ldquoReal-time implementation of fault-tolerant control systems with performance optimizationrdquo IEEETransactions on Industrial Electronics vol 64 no 5 pp 2402ndash2411 2014
[16] S Yin G Wang and H Karimi ldquoData-driven design of robustfault detection system for wind turbinesrdquoMechatronics 2013
[17] S Yin S X Ding A H A Sari and H Hao ldquoData-drivenmonitoring for stochastic systems and its application on batchprocessrdquo International Journal of Systems Science vol 44 no 7pp 1366ndash1376 2013
[18] S Yin S Ding A Haghani H Hao and P Zhang ldquoA com-parison study of basic datadriven fault diagnosis and process
monitoring methods on the benchmark Tennessee Eastmanprocessrdquo Journal of Process Control vol 22 no 9 pp 1567ndash15812012
[19] Y G Kao and C H Wang ldquoGlobal stability analysis forstochastic coupled reaction-diffusion systems on networksrdquoNonlinear Analysis Real World Applications vol 14 no 3 pp1457ndash1465 2013
[20] E K Boukas ldquoStabilization of stochastic singular nonlinearhybrid systemsrdquo Nonlinear Analysis Theory Methods amp Appli-cations vol 64 no 2 pp 217ndash228 2006
[21] E-K Boukas Control of Singular Systems with Random AbruptChanges Springer Berlin Germany 2008
[22] Y Ding H Zhu S Zhong and Y Zeng ldquoExponential mean-square stability of time-delay singular systems with Markovianswitching and nonlinear perturbationsrdquo Applied Mathematicsand Computation vol 219 no 4 pp 2350ndash2359 2012
[23] S Long S Zhong and Z Liu ldquoStochastic admissibility fora class of singular Markovian jump systems with mode-dependent time delaysrdquoAppliedMathematics and Computationvol 219 no 8 pp 4106ndash4117 2012
[24] Z Wu P Shi H Su and J Chu ldquoAsynchronous 1198682minus 119868infin
filtering for discrete-time stochastic Markov jump systems withrandomly occurred sensor nonlinearitiesrdquo Automatica vol 50no 1 pp 180ndash186 2014
[25] Z Wu P Shi H Su and J Chu ldquoStochastic synchronizationof Markovian jump neural networks with time-varying delayusing sampled-datardquo IEEE Transactions on Cybernetics vol 43no 6 pp 1796ndash1806 2013
[26] S Ma and E-K Boukas ldquoGuaranteed cost control of uncertaindiscrete-time singular Markov jump systems with indefinitequadratic costrdquo International Journal of Robust and NonlinearControl vol 21 no 9 pp 1031ndash1045 2011
[27] G Wang and Q Zhang ldquoRobust control of uncertain singularstochastic systems with Markovian switching via proportional-derivative state feedbackrdquo IET Control Theory amp Applicationsvol 6 no 8 pp 1089ndash1096 2012
[28] J Wang H Wang A Xue and R Lu ldquoDelay-dependent 119867infin
control for singular Markovian jump systems with time delayrdquoNonlinear Analysis Hybrid Systems vol 8 pp 1ndash12 2013
[29] L Wu X Su and P Shi ldquoSliding mode control with bounded1198712gain performance of Markovian jump singular time-delay
systemsrdquo Automatica vol 48 no 8 pp 1929ndash1933 2012[30] Z-GWu J H ParkH Su and J Chu ldquoStochastic stability anal-
ysis for discrete-time singular Markov jump systems with time-varying delay and piecewise-constant transition probabilitiesrdquoJournal of the Franklin Institute vol 349 no 9 pp 2889ndash29022012
[31] J Xiong J Lam H Gao and D W C Ho ldquoOn robust stabi-lization of Markovian jump systems with uncertain switchingprobabilitiesrdquo Automatica vol 41 no 5 pp 897ndash903 2005
[32] J Xiong and J Lam ldquoRobust 1198672control of Markovian jump
systems with uncertain switching probabilitiesrdquo InternationalJournal of Systems Science vol 40 no 3 pp 255ndash265 2009
[33] M Karan P Shi and C Y Kaya ldquoTransition probabilitybounds for the stochastic stability robustness of continuous-and discrete-timeMarkovian jump linear systemsrdquoAutomaticavol 42 no 12 pp 2159ndash2168 2006
[34] L Zhang and E-K Boukas ldquoStability and stabilization ofMarkovian jump linear systems with partly unknown transitionprobabilitiesrdquo Automatica vol 45 no 2 pp 463ndash468 2009
Abstract and Applied Analysis 11
[35] B Du J Lam Y Zou and Z Shu ldquoStability and stabilization forMarkovian jump time-delay systems with partially unknowntransition ratesrdquo IEEE Transactions on Circuits and Systems IRegular Papers vol 60 no 2 pp 341ndash351 2013
[36] L Zhang and J Lam ldquoNecessary and sufficient conditionsfor analysis and synthesis of Markov jump linear systemswith incomplete transition descriptionsrdquo IEEE Transactions onAutomatic Control vol 55 no 7 pp 1695ndash1701 2010
[37] Y Guo and F Zhu ldquoNew results on stability and stabilizationof Markovian jump systems with partly known transitionprobabilitiesrdquoMathematical Problems in Engineering vol 2012Article ID 869842 11 pages 2012
[38] J Lin S Fei and J Shen ldquoDelay-dependent 119867infin
filtering fordiscrete-time singular Markovian jump systems with time-varying delay and partially unknown transition probabilitiesrdquoSignal Processing vol 91 no 2 pp 277ndash289 2011
[39] J Tian Y Li J Zhao and S Zhong ldquoDelay-dependent stochasticstability criteria for Markovian jumping neural networks withmode-dependent time-varying delays and partially knowntransition ratesrdquo Applied Mathematics and Computation vol218 no 9 pp 5769ndash5781 2012
[40] Y Wei J Qiu H R Karimi and M Wang ldquoA new design 119867infin
filtering for contrinuous-time Markovian jump systems withtime-varying delay and partially accessible mode informationrdquoSignal Processing vol 93 pp 2392ndash2407 2013
[41] L Zhang E-K Boukas and J Lam ldquoAnalysis and synthesisof Markov jump linear systems with time-varying delays andpartially known transition probabilitiesrdquo IEEE Transactions onAutomatic Control vol 53 no 10 pp 2458ndash2464 2008
[42] L Zhang and E-K Boukas ldquoMode-dependent 119867infin
filteringfor discrete-time Markovian jump linear systems with partlyunknown transition probabilitiesrdquo Automatica vol 45 no 6pp 1462ndash1467 2009
[43] L Zhang and E-K Boukas ldquoMode-dependent 119867infin
controlfor discrete-time Markovian jump linear systems with partlyunknown transition probabilitiesrdquo International Journal ofRobust and Nonlinear Control vol 19 no 8 pp 868ndash883 2009
[44] Y Zhang Y He M Wu and J Zhang ldquoStabilization forMarkovian jump systems with partial information on transi-tion probability based on free-connection weighting matricesrdquoAutomatica vol 47 no 1 pp 79ndash84 2011
[45] G Zong D Yang L Hou and Q Wang ldquoRobust finite-time119867infincontrol for Markovian jump systems with partially known
transition probabilitiesrdquo Journal of the Franklin Institute vol350 no 6 pp 1562ndash1578 2013
[46] Y Ding H Zhu S Zhong and Y Zhang ldquo1198712minus 119871infin
filteringfor Markovian jump systems with time-varying delays andpartly unknown transition probabilitiesrdquo Communications inNonlinear Science and Numerical Simulation vol 17 no 7 pp3070ndash3081 2012
[47] Q Ma S Xu and Y Zou ldquoStability and synchronizationfor Markovian jump neural networks with partly unknowntransition probabilitiesrdquo Neurocomputing vol 74 pp 3403ndash3411 2011
[48] E Tian D Yue and G Wei ldquoRobust control for Markovianjump systems with partially known transition probabilities andnonlinearitiesrdquo Journal of the Franklin Institute vol 350 no 8pp 2069ndash2083 2013
[49] Y Guo and Z Wang ldquoStability of Markovian jump systemswith generally uncertain transition ratesrdquo Journal of the FranklinInstitute vol 350 no 9 pp 2826ndash2836 2013
[50] L Zhang ldquo119867infinestimation for discrete-time piecewise homoge-
neous Markov jump linear systemsrdquo Automatica vol 45 no 11pp 2570ndash2576 2009
[51] Z Wu H Su and J Chu ldquostate estimation for discreteMarkovian jumping neural networks with time delayrdquo Neuro-computing vol 73 pp 2247ndash2254 2010
[52] I R Petersen and A V Savkin Robust Kalman Filteringfor Signals and Systems with Large Uncertainties BirkhauserBoston Mass USA 1999
[53] B D O Anderson and J B Moore Optimal Filtering Prentice-Hall Upper Saddle River NJ USA 1979
[54] I R Petersen and D C McFarlane ldquoOptimal guaranteedcost control and filtering for uncertain linear systemsrdquo IEEETransactions on Automatic Control vol 39 no 9 pp 1971ndash19771994
[55] S Xu T Chen and J Lam ldquoRobust 119867infin filtering for uncertainMarkovian jump systems with mode-dependent time delaysrdquoIEEE Transactions on Automatic Control vol 48 no 5 pp 900ndash907 2003
[56] J H Kim D Oh and H Park ldquoGuaranteed cost and 119867infin
filtering for time-delay systemsrdquo in Proceeding of the AmericanControl Conference pp 4014ndash4018 IEEE Arlington Va USA2001
[57] L Xie ldquoOutput feedback119867infincontrol of systems with parameter
uncertaintyrdquo International Journal of Control vol 63 no 4 pp741ndash750 1996
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Abstract and Applied Analysis 3
119880119894
119896 then generally uncertain TR matrix (4) reduces to partly
uncertain TR matrix (6) as follows
[[[[
[
12058711
sdot sdot sdot
12058723
sdot sdot sdot 1205872119904
d
1205871199042
sdot sdot sdot
]]]]
]
(6)
Our results in this paper can be applicable to the generalMarkovian jump systems with bounded uncertain or partlyuncertain TR matrix
119860(120574(119905) 119905)119860119889(120574(119905) 119905)119862(120574(119905) 119905) and119862
119889(120574(119905) 119905) arematrix
functions of the random jumping process 120574(119905) To simplify thenotion the notation 119860
119894(119905) represents 119860(120574(119905) 119905) when 120574(119905) =
119894 For example 119860119889(120574(119905) 119905) is denoted by 119860
119889119894(119905) and so on
Further for each 120574(119905) = 119894 isin 119873 it is assumed that thematrices 119860
119894(119905) 119860
119889119894(119905) 119862119894(119905) and 119862
119889119894(119905) can be described by
the following form
119860119894(119905) = 119860
119894+ Δ119860119894(119905) 119860
119889119894(119905) = 119860
119889119894+ Δ119860119889119894(119905)
119862119894(119905) = 119862
119894+ Δ119862119894(119905) 119862
119889119894(119905) = 119862
119889119894+ Δ119862119889119894(119905)
(7)
where 119860119894 119860119889119894 119862119894are 119862
119889119894known real coefficient matrices
with appropriate dimensions Time-varying matricesΔ119860119894(119905) Δ119860
119889119894(119905) Δ119862
119894(119905) and Δ119862
119889119894(119905) represent norm-
bounded uncertainties and satisfy
[Δ119860119894(119905) Δ119860
119889119894(119905)
Δ119862119894(119905) Δ119862
119889119894(119905)
] = [1198721119894
1198722119894
]119865119894(119905) [1198731119894 1198732119894] (8)
where119872111989411987221198941198721119894 and119873
2119894are known constant real matri-
ces of appropriate dimensions which represent the structureof uncertainties and 119865
119894(119905) is an unknown matrix function
with Lebesgue measurable elements and satisfies 119865119894(119905)119865119879
119894(119905) le
119868Further for convenience we assume that the system has
the same dimension at each mode and the Markov processis irreducible Consider the following nominal unforceddescriptor time-delay system
119864 (119905) = 119860119894119909 (119905) + 119860
119889119894119909 (119905 minus 119889)
119909 (119905) = 120593 (119905) forall119905 isin [minus119889 0]
(9)
Let 1199090 1199030 and 119909(119905 120593 119903
0) be the initial state initial mode
and the corresponding solution of the system (9) at time 119905respectively
Definition 5 System (9) is said to be stochastically stable iffor all 120593(119905) isin 119871
2[minus119889 0] and initial mode 119903
0isin 119873 there exists
a matrix119872 gt 0 such that
119864int
infin
0
1003817100381710038171003817119909 (119905 120593 1199030)10038171003817100381710038172
119889119905 | 1199030 119909 (119905) = 120593 (119905) 119905 isin [minus119889 0]
le 119909119879
01198721199090
(10)
The following definition can be regarded as an extensionof the definition in [2]
Definition 6 (1) System (9) is said to be regular if det(119904E minus
119860119894) 119894 = 1 2 119904 are not identically zero(2) System (9) is said to be impulse free if deg(det(119904119864 minus
119860119894)) = rank 119864
119894 119894 = 1 2 119904
(3) System (9) is said to be admissible if it is regularimpulse free and stochastically stable
The linearmemoryless observer under consideration is asfollows
119864 119909 (119905) = 1198701119894119909 (119905) + 119870
2119894119910 (119905)
1199090= 0 119903 (0) = 119903
0
(11)
where 119909(119905) isin 119877119899 is the observer state and the constant
matrices1198701119894and119870
2119894are observer parameters to be designed
Denote the error state 119890(119905) = 119909(119905) minus 119909(119905) and theaugmented state vector 119909
119891= [119909119879(119905) 119890119879(119905)]119879
Let 119909(119905) = 119871119890(119905)
represent the output of the error states where 119871 is a knownconstant matrix Define
119860119891119894= [
119860119894
0
119860119894minus 1198701119894minus 11987021198941198621198941198701119894
]
119860119891119889119894
= [119860119889119894
0
119860119889119894minus 1198702119894119862119889119894
0] 119864
119891= [
119864 0
0 119864]
119872119891119894= 1198721198911119894
= [1198721119894
1198721119894minus 11987021198941198722119894
] 119873119891119894= [1198731119894 0]
Δ119860119891119894= 119872119891119894119865119894(119905)119873119891119894 119873
1198911119894= [1198732119894 0]
Δ119860119891119889119894
= 1198721198911119894
119865119894(119905)1198731198911119894
119862119891= [0 119871]
(12)
and combine (1) and (11) then we derive the augmentedsystems as follows
119864119891119891(119905) = (119860
119891119894+ Δ119860119891119894) 119909119891(119905)
+ (119860119891119889119894
+ Δ119860119891119889119894
) 119909119891(119905 minus 119889)
119911 (119905) = 119862119891119909119891(119905)
1199091198910
(119905) = [120593119879(119905) 120593119879(119905)]119879
forall119905 isin [minus119889 0]
(13)
Similar to [5] it is also assumed in this paper that for all 120589 isin[minus119889 0] there exists a scalar ℎ gt 0 such that 119909
119891(119905 + 120589) le
ℎ119909119891(119905)Associated with system (13) is the cost function
J = Eint
infin
0
119911119879(119905) 119911 (119905) 119889119905 (14)
Definition 7 Consider the augmented system (13) if thereexist the observer parameters 119870
1119894 1198702119894and a positive scalar
Jlowast for all uncertainties such that the augmented system(13) is robust stochastically stable and the value of the costfunction (14) satisfies J le Jlowast then Jlowast is said to be arobust guaranteed cost and observer (11) is said to be a robustguaranteed cost observer for system (1) with (4)
Problem 8 (robust guaranteed cost observer problem for aclass of linear singular Markovian jump time-delay systems
4 Abstract and Applied Analysis
with generally incomplete transition probability) Given sys-tem (1) with GUTRMatrix (4) can we determine an observer(11) with parameters 119870
1119894and 119870
2119894such that the observer is a
robust guaranteed cost observer for system (1) with GUTRMatrix (4)
Lemma 9 Given any real number 120576 and any matrix Q thematrix inequality 120576(119876 + 119876
119879) le 1205762119879 + 119876119879
minus1119876119879 holds for any
matrix 119879 gt 0
3 Main Results
Theorem 10 Consider the augmented system (13)with GUTRMatrix (4) and the cost function (14) Then the robust guar-anteed cost observer (11) with parameters 119870
1119894and 119870
2119894can
be designed if there exist matrices 119875119894 1198701119894 and 119870
2119894 119894 =
1 2 119904 and symmetric positive definite matrix Q satisfyingthe following LMIs respectively
Case 1 If 119894 notin 119880119894
119896and 119880
119894
119896= 119896119894
1 119896
119894
119898 there exist a set of
symmetric positive definite matrices 119879119894119895isin R119899times119899 (119894 notin 119880
119894
119896 119895 isin
119880119894
119896) such that
119864119879
119891119875119894= 119875119879
119894119864119891ge 0 (15)
[[
[
Π119894+ 119862119879
119891119862119891
119875119894(119860119891119889119894
+ Δ119860119891119889119894
) 1
(119860119891119889119894
+ Δ119860119891119889119894
)119879
119875119894
minus119876 0
lowast lowast 2
]]
]
lt 0 (16)
119875119894minus 119875119895ge 0 forall119895 isin 119880
119894
119906119896 119895 = 119894 (17)
Case 2 If 119894 isin 119880119894
119896 119880119894119896= 119896119894
1 119896
119894
119898 and 119880119894
119906119896= 0 there exist a
set of symmetric positive definite matrices 119881119894119895119897
isin R119899times119899 (119894 119895 isin
119880119894
119896 119897 isin 119880
119894
119906119896) such that
119864119879
119891119875119894= 119875119879
119894119864119891ge 0 (18)
[[
[
Ω119894+ 119862119879
119891119862119891
119875119894(119860119891119889119894
+ Δ119860119891119889119894
) 1
(119860119891119889119894
+ Δ119860119891119889119894
)119879
119875119894
minus119876 0
lowast lowast 2
]]
]
lt 0 (19)
Case 3 If 119894 isin 119880119894
119896and 119880
119894
119906119896= 0 there exist a set of symmetric
positive definite matrices119882119894119895isin R119899times119899 (119894 119895 isin 119880
119894
119896) such that
119864119879
119891119875119894= 119875119879
119894119864119891ge 0 (20)
[[
[
Δ119894+ 119862119879
119891119862119891
119875119894(119860119891119889119894
+ Δ119860119891119889119894
) 1
(119860119891119889119894
+ Δ119860119891119889119894
)119879
119875119894
minus119876 0
lowast lowast 2
]]
]
lt 0 (21)
where
Π119894= (119860119891119894+ Δ119860119891119894)119879
119875119894+ 119875119894(119860119891119894+ Δ119860119891119894)
+ 119876 + sum
119895isin119880119894
119896
119894119895119864119879
119891(119875119895minus 119875119894) + sum
119895isin119880119894
119896
1
41205902
119894119895119879119894119895
Ω119894= (119860119891119894+ Δ119860119891119894)119879
119875119894+ 119875119894(119860119891119894+ Δ119860119891119894)
+ 119876 + sum
119895isin119880119894
119896
119894119895119864119879
119891(119875119895minus 119875119897) + sum
119895isin119880119894
119896
1
41205902
119894119895119881119894119895119897
Δ119894= (119860119891119894+ Δ119860119891119894)119879
119875119894+ 119875119894(119860119891119894+ Δ119860119891119894)
+ 119876 + sum
119895isinS119895 = 119894
119894119895119864119879
119891(119875119895minus 119875119894) + sum
119895isinS119895 = 119894
1
41205902
119894119895119882119894119895
1= [119864119879
119891(119875119894119896119894
1
minus 119875119894) 119864119879
119891(119875119894119896119894
2
minus 119875119894) 119864
119879
119891(119875119894119896119894
119898
minus 119875119894)]
2= diag minus119879
119894119896119894
1
minus119879119894119896119894
119898
1= [119864119879
119891(119875119896119894
1
minus 119875119897) 119864119879
119891(119875119896119894
2
minus 119875119897) 119864
119879
119891(119875119896119894
119898
minus 119875119897)]
2= diag minus119881
119894119896119894
1119897 minus119881
119894119896119894
119898119897
1= [119864119879
119891(1198751minus 119875119894) 119864
119879
119891(119875119894minus1
minus 119875119894)
119864119879
119891(119875119894+1
minus 119875119894) 119864
119879
119891(119875119904minus 119875119894)]
2= diag minus119882
1198941 minus119882
119894119904
(22)
Proof According to Definition 2 and Theorem 1 in [2] wecan derive from (15)ndash(21) that system (13) is regular andimpulse free Let the mode at time 119905 be 119894 and consider thefollowing Lyapunov function with respect to the augmentedsystem (13)
119881(119909119891(119905) 120574 (119905) = 119894) = 119909
119879
119891(119905) 119864119879
119891119875119894119909119891(119905)
+ int
119905
119905minus119889
119909119879
119891(119904) 119876119909
119891(119904) 119889119905
(23)
where 119876 is the symmetric positive definite matrix to be cho-sen and 119875
119894is a matrix satisfying (15)ndash(21)The weak infinites-
imal operatorL of the stochastic process 120574(119905) 119909119891(119905) 119905 ge 0
is presented by
L119881(119909119891(119905) 120574 (119905) = 119894)
= limΔrarr0
1
Δ[119864119891119881 (119909 (119905 + Δ) 120574 (119905 + Δ)) 119909 (119905) 120574 (119905) = 119894
minus119881 (119909 (119905) 120574 (119905) = 119894) ]
= 119909119879
119891(119905) [
[
(119860119891119894+ Δ119860119891119894)119879
119875119894+ 119875119894(119860119891119894+ Δ119860119891119894)
+
119904
sum
119895=1
120587119894119895119864119879
119891119875119895+ 119876]
]
119909119891(119905)
+ 2119909119879
119891(119905) 119875119894(1198601198911119894
+ Δ1198601198911119894
) 119909119891(119905 minus 119889)
minus 119909119879
119891(119905 minus 119889)119876119909
119891(119905 minus 119889)
(24)
Abstract and Applied Analysis 5
Case 1 (119894 notin 119880119894
119896) Note that in this case sum
119895isin119880119894
119906119896119895 = 119894
120587119894119895
=
minussum119895isinU119894119896119895 = 119894
120587119894119895minus 120587119894119894and 120587
119894119895ge 0 119895 isin 119880
119894
119906119896 119895 = 119894 then from
(24) we have
119909119879
119891(119905) [
[
119904
sum
119895=1
120587119894119895119864119879
119891119875119895]
]
119909119891(119905)
= 119909119879
119891(119905) [
[
sum
119895isin119880119894
119896
120587119894119895119864119879
119891119875119895+ sum
119895isin119880119894
119906119896119895 = 119894
120587119894119895119864119879
119891119875119895+ 120587119894119894119864119879
119891119875119895]
]
119909119891(119905)
= 119909119879
119891(119905) [
[
sum
119895isin119880119894
119896
120587119894119895119864119879
119891119875119895+ (minus120587
119894119894minus sum
119895isin119880119894
119896
120587119894119895)119864119879
119891119875119894
+120587119894119894119864119879
119891119875119894]
]
119909119891(119905)
= 119909119879
119891(119905) 119864119879
119891[
[
sum
119895isin119880119894
119896
120587119894119895(119875119895minus 119875119894)]
]
119909119891(119905)
= 119909119879
119891(119905) 119864119879
119891[
[
sum
119895isin119880119894
119896
119894119895(119875119895minus 119875119894) + sum
119895isin119880119894
119896
Δ119894119895(119875119895minus 119875119894)]
]
119909119891(119905)
(25)
On the other hand in view of Lemma 9 we have
sum
119895isin119880119894
119896
Δ119894119895119864119879
119891(119875119895minus 119875119894)
= sum
119895isin119880119894
119896
[1
2Δ119894119895119864119879
119891(119875119895minus 119875119894) +
1
2Δ119894119895119864119879
119891(119875119895minus 119875119894)]
le sum
119895isin119880119894
119896
[(1
2Δ119894119895)
2
119879119894119895+ 119864119879
119891(119875119895minus 119875119894) 119879minus1
119894119895(119875119895minus 119875119894) 119864119891]
le sum
119895isin119880119894
119896
[1
41205902
119894119895119879119894119895+ 119864119879
119891(119875119895minus 119875119894) 119879minus1
119894119895(119875119895minus 119875119894) 119864119891]
(26)
Case 2 (119894 isin 119880119894
119896and119880119894
119906119896= 0) Because of119880119894
119896= 119896119894
1 119896
119894
119898 and
119880119894
119906119896= 119906119894
1 119906
119894
119904minus119898 there must be 119897 isin 119880
119894
119906119896so that 119864119879
119891119875119897ge
119864119879
119891119875119895(for all 119895 isin 119880
119894
119906119896)
119909119879
119891(119905) [
[
119904
sum
119895=1
120587119894119895119864119879
119891119875119895]
]
119909119891(119905)
le 119909119879
119891(119905) [
[
sum
119895isin119880119894
119896
120587119894119895119864119879
119891119875119895minus ( sum
119895isin119880119894
119896119895 = 119894
120587119894119895)119864119879
119891119875119897]
]
119909119891(119905)
= 119909119879
119891(119905) 119864119879
119891[
[
sum
119895isin119880119894
119896
120587119894119895(119875119895minus 119875119897)]
]
119909119891(119905)
= 119909119879
119891(119905) 119864119879
119891[
[
sum
119895isin119880119894
119896
119894119895(119875119895minus 119875119897) + sum
119895isin119880119894
119896
Δ119894119895(119875119895minus 119875119897)]
]
119909119891(119905)
(27)
By using Lemma 9 we have
sum
119895isin119880119894
119896
Δ119894119895119864119879
119891(119875119895minus 119875119897)
= sum
119895isin119880119894
119896
[1
2Δ119894119895119864119879
119891(119875119895minus 119875119897) +
1
2Δ119894119895119864119879
119891(119875119895minus 119875119897)]
le sum
119895isin119880119894
119896
[(1
2Δ119894119895)
2
119881119894119895119897+ 119864119879
119891(119875119895minus 119875119897)119881minus1
119894119895119897(119875119895minus 119875119897) 119864119879
119891]
le sum
119895isin119880119894
119896
[1
41205902
119894119895119881119894119895119897+ 119864119879
119891(119875119895minus 119875119897)119881minus1
119894119895119897(119875119895minus 119875119897) 119864119879
119891]
(28)
Case 3 (119894 isin 119880119894
119896and 119880
119894
119906119896= 0) Consider
119909119879
119891(119905) [
[
119904
sum
119895=1
120587119894119895119864119879
119891119875119895]
]
119909119891(119905)
= 119909119879
119891(119905) 119864119879
119891[
[
119904
sum
119895=1119895 = 119894
120587119894119895(119875119895minus 119875119894)]
]
119909119891(119905)
= 119909119879
119891(119905) 119864119879
119891[
[
119904
sum
119895=1119895 = 119894
119894119895(119875119895minus 119875119894)
+
119904
sum
119895=1119895 = 119894
Δ119894119895(119875119895minus 119875119894)]
]
119909119891(119905)
(29)
Case 1 Substituting (25) and (26) into (24) it results in
L119881 le Λ119879(119905) Φ (119894) Λ (119905) (30)
where Λ119879(119905) = [119909119879
119891(119905) 119909119879
119891(119905 minus 119889)] and
Φ119894=[[
[
Π119894+ 119862119879
119891119862119891
119875119894(119860119891119889119894
+ Δ119860119891119889119894
) 1
(119860119891119889119894
+ Δ119860119891119889119894
)119879
119875119894
minus119876 0
lowast lowast 2
]]
]
(31)
Case 2 Substituting (27) and (28) into (24) it results in
L119881 le Λ119879(119905) Ψ (119894) Λ (119905) (32)
where Λ119879(119905) = [119909119879
119891(119905) 119909119879
119891(119905 minus 119889)] and
Ψ119894=[[
[
Ω119894+ 119862119879
119891119862119891
119875119894(119860119891119889119894
+ Δ119860119891119889119894
) 1
(119860119891119889119894
+ Δ119860119891119889119894
)119879
119875119894
minus119876 0
lowast lowast 2
]]
]
(33)
6 Abstract and Applied Analysis
Case 3 Substituting (29) into (24) we get
L119881 le Λ119879(119905) Γ (119894) Λ (119905) (34)
where Λ119879(119905) = [119909119879
119891(119905) 119909119879
119891(119905 minus 119889)] and
Γ119894=[[
[
Δ119894+ 119862119879
119891119862119891
119875119894(119860119891119889119894
+ Δ119860119891119889119894
) 1
(119860119891119889119894
+ Δ119860119891119889119894
)119879
119875119894
minus119876 0
lowast lowast 2
]]
]
(35)
Similar to [5] usingDynkinrsquos formula we drive for each 119894 isin 119873
lim119879rarrinfin
Eint
119879
0
119909119879
119891(119905) 119909119891(119905) 119889119905 | 120593
119891 1205740= 119894 le 119909
119879
11989101198721199091198910 (36)
By Definition 5 it is easy to see that the augmented system(13) is stochastically stable Furthermore from (16) (19) and(21) we have
L119881 le minus119909119879
119891(119905) 119862119879
119891119862119891119909119891(119905) lt 0 (37)
On the other hand we have
J = Eint
infin
0
119909119879
119891(119905) 119862119879
119891119862119891119909119891(119905) 119889119905 lt minusint
infin
0
L119881119889119905
= minus E lim119905rarrinfin
119881 (119909 (119905) 120574 (119905)) + 119881 (1199090 1205740)
(38)
As the augmented system (13) is stochastically stable itfollows from (38) that 119869 lt 119881(119909
1198910 1199030) From Definition 7 it
is concluded that a robust guaranteed cost for the augmentedsystem (13) can be given by 119869
lowast= 119909119879
1198910(119905)119864119879
1198911199030119875(1199030)1199091198910
+
int0
minus119889119909119879
119891(119905)119876119909
119891(119905)119889119905
In the following based on the above sufficient conditionthe design of robust guaranteed cost observers can be turnedinto the solvability of a system of LMIs
Theorem 11 Consider system (13)with GUTRMatrix (4) andthe cost function (14) If there exist matrices 119884
1119894and 119884
2119894 119894 =
1 2 119904 positive scalars 120576119894 119894 = 1 2 119904 symmetric positive
definite matrix Q and the full rank matrices 1198752119894 and matrices
119875119894= diag(119875
1119894 1198752119894) 119894 = 1 2 119904 satisfying the following
LMIs respectively
Case 1 If 119894 notin 119880119894
119896and 119880
119894
119896= 119896119894
1 119896
119894
119898 a set of positive
definite matrices 119879119894119895isin R119899times119899 (119894 notin 119880
119894
119896 119895 isin 119880
119894
119896) exist such that
119864119879
119891119875119894= 119875119879
119894119864119891ge 0 (39)
[[[[[[[[[
[
1206011119894
1206012119894
1198731
1206013119894
120601119879
2119894minus119876 0 0
119873119879
10 119873
20
120601119879
31198940 0 minus120576
119894119868
]]]]]]]]]
]
lt 0 (40)
119875119894minus 119875119895ge 0 forall119895 isin 119880
119894
119906119896 119895 = 119894 (41)
Case 2 If 119894 isin 119880119894
119896(119880119894
119896= 119896119894
1 119896
119894
119898) and 119880
119894
119906119896= 0 a set of
positive definite matrices119881119894119895119897isin R119899times119899 (119894 119895 isin 119880
119894
119896 119897 isin 119880
119894
119906119896) exist
such that
119864119879
119891119875119894= 119875119879
119894119864119891ge 0 (42)
[[[[[[[[[
[
1205931119894
1205932119894
1198721
1205933119894
120593119879
2119894minus119876 0 0
119872119879
10 119872
20
120593119879
31198940 0 minus120576
119894119868
]]]]]]]]]
]
lt 0 (43)
Case 3 If 119894 isin 119880119894
119896and 119880
119894
119906119896= 0 a set of positive definite
matrices119882119894119895isin R119899times119899 (119894 119895 isin 119880
119894
119896) exist such that
119864119879
119891119894119875119894= 119875119879
119894119864119891119894ge 0 (44)
[[[[[[[[[
[
1205951119894
1205952119894
1198711
1205953119894
120595119879
2119894minus119876 0 0
119871119879
10 1198712
0
120595119879
31198940 0 minus120576
119894119868
]]]]]]]]]
]
lt 0 (45)
where
1206011119894= 1205931119894= 1205951119894
= [1198751119894119860119894+ 119860119879
1198941198751119894
119860119879
1198941198752119894minus 119884119879
1119894minus 119862119879
119894119884119879
2119894
1198752119894119860119894minus 1198841119894minus 1198842119894119862119894
119884119879
1119894+ 1198841119894
]
+ 119876 + 119862119879
119891119862119891+ sum
119895isin119880119894
119896
119894119895119864119879
119891(119875119895minus 119875119894) + sum
119895isin119880119894
119896
1
41205902
119894119895119879119894119895
1206012119894= 1205932119894= 1205952119894= [
1198751119894119860 i 0
1198752119894119860119894minus 1198841119894minus 11988421198941198621198940]
1206013119894= 1205933119894= 1205953119894= [
11987511198941198721119894
11987521198941198721119894minus 11988411198941198721119894minus 11988421198941198722119894
]
1198731= [119864119879
119891(119875119896119894
1
minus 119875119894) 119864119879
119891(119875119896119894
2
minus 119875119894) 119864
119879
119891(119875119896119894
119898
minus 119875119894)]
1198732= diag minus119879
119894119896119894
1
minus119879119894119896119894
119898
1198721= [119864119879
119891(119875119896119894
1
minus 119875119894) 119864119879
119891(119875119896119894
2
minus 119875119894) 119864
119879
119891(119875119896119894
119898
minus 119875119894)]
1198722= diag minus119881
119894119896119894
1119897 minus119881
119894119896119894
119898119897
1198711= [119864119879
119891(1198751minus 119875119894) 119864
119879
119891(119875119894minus1
minus 119875119894)
119864119879
119891(119875119894+1
minus 119875119894) 119864
119879
119891(119875119904minus 119875119894)]
1198712= diag minus119882
1198941 minus119882
119894119904
(46)
Abstract and Applied Analysis 7
Then a suitable robust guaranteed cost observer in theform of (11) has parameters as follows
1198701119894= 119875minus1
11198941198841119894 119870
2119894= 119875minus1
21198941198842119894
(47)
and 119869lowast is a robust guaranteed cost for system (13) with GUTRMatrix (4)
Proof Define
1198601
119894=
[[[[
[
119860119879
119891119894119875119894+ 119875119894119860119891119894+ 119876 + sum
119895isin119880119894
119896
119894119895(119875119895minus 119875119894) + sum
119895isin119880119894
119896
1
41205902
119894119895119879119894119895+ 119862119879
119891119862119891
119875119894119860119891119889119894
1198731
119860119879
119891119889119894119875119894
minus119876 0
lowast lowast 1198732
]]]]
]
(48)
1198602
119894=
[[[[
[
119860119879
119891119894119875119894+ 119875119894119860119891119894+ 119876 + sum
119895isin119880119894
119896
119894119895(119875119895minus 119875119894) + sum
119895isin119880119894
119896
1
41205902
119894119895119881119894119895119897+ 119862119879
119891119862119891
119875119894119860119891119889119894
1198721
119860119879
119891119889119894119875119894
minus119876 0
lowast lowast 1198722
]]]]
]
(49)
1198603
119894=
[[[[
[
119860119879
119891119894119875119894+ 119875119894119860119891119894+ 119876 + sum
119895isin119880119894
119896
119894119895(119875119895minus 119875119894) + sum
119895isin119880119894
119896
1
41205902
119894119895119882119894119895+ 119862119879
119891119862119891
119875119894119860119891119889119894
1198711
119860119879
119891119889119894119875119894
minus119876 0
lowast lowast 1198712
]]]]
]
lt 0 (50)
Then (16) is equivalent to
1198601
119894+ [
[
119875119894119872119891119894
0
0
]
]
119865119894[119873119891119894 1198731198911119894 0]
+ [119873119891119894 1198651198911119894 0]119879
119865119879
119894[
[
119875119894119872119891119894
0
0
]
]
119879
lt 0
(51)
Then (19) is equivalent to
1198602
119894+ [
[
119875119894119872119891119894
0
0
]
]
119865119894[119873119891119894 1198731198911119894 0]
+ [119873119891119894 1198651198911119894 0]119879
119865119879
119894[
[
119875119894119872119891119894
0
0
]
]
119879
lt 0
(52)
Then (21) is equivalent to
1198603
119894+ [
[
119875119894119872119891119894
0
0
]
]
119865119894[119873119891119894 1198731198911119894 0]
+ [119873119891119894 1198651198911119894 0]119879
119865119879
119894[
[
119875119894119872119891119894
0
0
]
]
119879
lt 0
(53)
By applying Lemma 24 in [57] (50) (51) and (52) hold forall uncertainties119865i satisfying119865
119879
119894119865119894lt 119868 if and only if there exist
positive scalars 120576119894 119894 = 1 2 119904 such that
1198601
119894+ 120576minus1
119894[
[
119875119894119872119891119894
0
0
]
]
[
[
119875119894119872119891119894
0
0
]
]
119879
+ 120576119894[119873119891119894 1198651198911119894 0]
119879
[119873119891119894 1198651198911119894 0] lt 0
1198602
119894+ 120576minus1
119894[
[
119875119894119872119891119894
0
0
]
]
[
[
119875119894119872119891119894
0
0
]
]
119879
+ 120576119894[119873119891119894 1198651198911119894 0]
119879
[119873119891119894 1198651198911119894 0] lt 0
1198603
119894+ 120576minus1
119894[
[
119875119894119872119891119894
0
0
]
]
[
[
119875119894119872119891119894
0
0
]
]
119879
+ 120576119894[119873119891119894 1198651198911119894 0]
119879
[119873119891119894 1198651198911119894 0] lt 0
(54)
Let 119875119894= diag(119875
1119894 1198752119894) and using (47) we can conclude from
Schur complement results that the above matrix inequalitiesare equivalent to the coupled LMIs (40) (43) and (45)It further follows from Theorem 10 that 119869
lowast is a robustguaranteed cost for system (13) with (4)
Remark 12 The solution of LMIs (39)ndash(45) parameterizesthe set of the proposed robust guaranteed cost observersThis parameterized representation can be used to design theguaranteed cost observer with some additional performanceconstraints By applying the methods in [14] the suboptimal
8 Abstract and Applied Analysis
guaranteed cost observer can be determined by solving acertain optimization problem This is the following theorem
Theorem 13 Consider system (13)with GUTRMatrix (4) andthe cost function (14) and suppose that the initial conditions 119903
0
and 1199091198910
are known if the following optimization problem
min1198761198751119894 1198752119894 120576119894 1198841119894 and 1198842119894
119869lowast
st LMIs (39) ndash (45)(55)
has a solution 119876 1198751119894 1198752119894 120576119894 1198841119894 and 119884
2119894 119894 = 1 2 119904
then the observer (11) is a suboptimal guaranteed costobserver for system (1) where 119869
lowast= 119909119879
1198910119864119879
1198911199030119875(1199030)1199091198910
+
tr(int0minus119889
1199091198910(119905)1199091198910(119905)119909119879
1198910119889119905119876)
Proof It follows from Theorem 11 that the observer (11)constructed in terms of the solution 119876 119875
1119894 1198752119894 120576119894 1198841119894 and
1198842119894 119894 = 1 2 119904 is a robust guaranteed cost observer By
noting that
int
0
minus119889
119909119879
1198910(119905) 119876119909
1198910(119905) 119889119905 = int
0
minus119889
tr (1199091198791198910
(119905) 1198761199091198910
(119905)) 119889119905
= tr(int0
minus119889
119909119879
1198910(119905) 1199091198910
(119905) 119889119905119876)
(56)
it follows that the suboptimal guaranteed cost observerproblem is turned into the minimization problem (55)
Remark 14 Theorem 13 gives the suboptimal guaranteed costobserver conditions of a class of linear Markovian jump-ing time-delay systems with generally incomplete transitionprobability and LMI constraints which can be easily solvedby the LMI toolbox in MATLAB
4 Numerical Example
In this section a numerical example is presented todemonstrate the effectiveness of the method mentioned inTheorem 11 Consider a 2-dimensional system (1) with 3Markovian switching modes In this numerical example thesingular system matrix is set as 119864 = [ 1 0
0 0] and the 3-mode
transition rate matrix is Λ = [minus32
2
15 21 minus36
] where Δ11 Δ31
isin
[minus015 015] Δ23 Δ33
isin [minus012 012] and Δ32
isin [minus01 01]The other system matrices are as follows
For mode 119894 = 1 there are
1198601= [
minus32 065
1 02] 119860
1198891= [
02 05
1 minus068]
1198621= [
[
12 065
minus65 19
minus021 minus18
]
]
1198621198891
= [
[
minus36 minus105
21 096
021 minus086
]
]
11987211
= [minus02
08] 119872
21= [
[
025
0875
minus2
]
]
11987311
= [minus12 31] 11987321
= [minus069 minus42]
(57)
For mode 119894 = 2 there are
1198602= [
minus1 6
2 minus36] 119860
1198892= [
minus31 minus16
3 075]
1198622= [
[
9 minus25
035 minus2
36 minus18
]
]
1198621198892
= [
[
089 minus6
minus12 09
minus24 6
]
]
11987212
= [23
minus4] 119872
22= [
[
075
minus36
25
]
]
11987312
= [minus72 minus6] 11987322
= [1 2]
(58)
For mode 119894 = 3 there are
1198603= [
minus106 29
minus03 36] 119860
1198893= [
minus56 minus12
minus3 45]
1198623= [
[
minus3 minus036
015 minus18
09 minus5
]
]
1198621198893
= [
[
minus165 5
minus12 265
minus098 minus56
]
]
11987213
= [minus82
minus03] 119872
23= [
[
minus052
25
minus36
]
]
11987313
= [105 minus5] 11987323
= [minus72 minus126]
(59)
Then we set the error state matrix 119871 = [ minus45 062 minus6
] and thepositive scalars in Theorem 11 are 120576
1= 02 120576
2= 015 120576
3=
032 According to the definitions of augmented statematricesin (12) we can easily obtain the following parameter matricesin Theorem 11 by MATLAB
11988411
= [minus84521006 00127
00127 84509001]
11988421
= [002 01291 01435
minus23520 minus04080 minus08106]
11988412
= [minus170991 269626
269626 minus246750]
11988422
= [minus200744 minus132941 529388
216893 180120 minus506693]
11988413
= [minus6751329 224456
224456 minus8976976]
11988423
= [minus136021 minus1461726 540500
minus34324 minus195125 minus103068]
1198751=[[[
[
63029 0 0 0
0 48620 0 0
0 0 22914 0
0 0 0 03169
]]]
]
Abstract and Applied Analysis 9
1198752=[[[
[
08914 0 0 0
0 12505 0 0
0 0 73629 0
0 0 0 30056
]]]
]
1198753=[[[
[
30265 0 0 0
0 02156 0 0
0 0 08965 0
0 0 0 10002
]]]
]
119876 =[[[
[
05000 0 0 0
0 05001 0 0
0 0 05001 0
0 0 0 05001
]]]
]
11987911
=[[[
[
34173214 minus8707765 0 0
minus8707765 4167216 0 0
0 0 22263598 minus3207456
0 0 minus3207456 12263101
]]]
]
11987923
=[[[
[
37753231 minus27999330 0 0
minus27999330 28107685 0 0
0 0 106907366 minus107432750
0 0 minus107432750 108555053
]]]
]
11987931
=[[[
[
9518504 minus5399245 0 0
minus5399245 8962029 0 0
0 0 14773012 minus2077540
0 0 minus2077540 14791256
]]]
]
11987932
=[[[
[
21617695 minus12094164 0 0
minus12094164 20371205 0 0
0 0 14786 minus09283
0 0 minus09283 14794
]]]
]
11987933
=[[[
[
14939313 minus8398780 0 0
minus8398780 14073689 0 0
0 0 1478123 minus1336452
0 0 minus1336452 2459347
]]]
]
11988111
=[[[
[
16650 0 0 0
0 16650 0 0
0 0 16650 0
0 0 0 16650
]]]
]
11988231
=[[[
[
15426 0 0 0
0 16650 0 0
0 0 16662 0
0 0 0 16650
]]]
]
11988232
=[[[
[
15428 0 0 0
0 16650 0 0
0 0 16622 0
0 0 0 16650
]]]
]
(60)
Therefore we can design a linear memoryless observer as(11) with the constant matrices
11987011
= 119875minus1
1111988411
= [minus13409860 00020
00026 17381530]
11987021
= 119875minus1
2111988421
= [00087 00563 00626
minus74219 minus12875 minus25579]
11987012
= 119875minus1
1211988412
= [minus191823 302475
215615 minus197321]
11987022
= 119875minus1
2211988422
= [minus27264 minus18056 71899
72163 59928 minus168583]
11987013
= 119875minus1
1311988413
= [minus2231402 74186
1041076 minus41637180]
11987023
= 119875minus1
2311988423
= [minus151724 minus1630481 602900
minus34317 minus195086 minus103047]
(61)
Finally the observer (11) with the above parameter matri-ces for this numerical example is a suboptimal guaranteedcost observer byTheorems 11 and 13
5 Conclusions
In this paper the robust guaranteed cost observer problemfor a class of uncertain descriptor time-delay systems withMarkovian jumping parameters and generally uncertaintransition rates is studied by using LMI method In thisGUTR singular model each transition rate can be com-pletely unknown or only its estimate value is known Theparameterrsquos uncertainty is time varying and is assumed tobe norm-bounded Memoryless guaranteed cost observersare designed in terms of a set of linear coupled matrixinequalities The suboptimal guaranteed cost observer isdesigned by solving a certain optimization problem Ourresults can be applicable to the general Markovian jumpsystems with bounded uncertain or partly uncertain TRmatrix
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This research is supported by the National NaturalScience Foundations of China (10701020 1097102260973048 61272077 60974025 60673101 and 60939003)NCET-08-0755 National 863 Plan Project (2008AA04Z401 2009AA043404) SDPW IMZQWH010016the Natural Science Foundation of Shandong Province (noZR2012FM006) the Scientific and Technological Projectof Shandong Province (no 2007GG3WZ04016) and theNatural Science Foundation of Guangxi Autonomous Region(no 2012GXNSFBA053003)
10 Abstract and Applied Analysis
References
[1] L Dai Singular Control Systems Springer Berlin Germany1989
[2] S Xu P V Dooren R Stefan and J Lam ldquoRobust stability andstabilization for singular systemswith state delay and parameteruncertaintyrdquo IEEE Transactions on Automatic Control vol 47no 7 pp 1122ndash1128 2002
[3] M Fu and G Duan ldquoRobust guaranteed cost observer foruncertain descriptor time-delay systems withMarkovian jump-ing parametersrdquo Acta Automatica Sinica vol 31 pp 479ndash4832005
[4] Y Q Xia E-K Boukas P Shi and J H Zhang ldquoStabilityand stabilization of continuous-time singular hybrid systemsrdquoAutomatica vol 45 no 6 pp 1504ndash1509 2009
[5] S Xu and J Lam Robust Control and Filtering of SingularSystems Springer Berlin Germany 2006
[6] Y-Y Cao and J Lam ldquoRobust 119867infin
control of uncertainMarkovian jump systems with time-delayrdquo IEEE Transactionson Automatic Control vol 45 no 1 pp 77ndash83 2000
[7] Y Li and Y Kao ldquoStability of stochastic reaction-diffusion sys-tems with Markovian switching and impulsive perturbationsrdquoMathematical Problems in Engineering vol 2012 Article ID429568 13 pages 2012
[8] X Mao and C Yuan Stochastic Differential Equations withMarkovian Switching Imperial College Press London UK2006
[9] E-K Boukas Stochastic Switching Systems Analysis and DesignBirkhauser Basel Switzerland 2005
[10] Y Kao C Wang and L Zhang ldquoDelay-dependent exponentialstability of impulsive markovian jumping Cohen-Grossbergneural networks with reaction-diffusion and mixed delaysrdquoNeural Processing Letters vol 38 no 3 pp 321ndash346 2013
[11] Y-G Kao J-F Guo C-H Wang and X-Q Sun ldquoDelay-dependent robust exponential stability of Markovian jump-ing reaction-diffusion Cohen-Grossberg neural networks withmixed delaysrdquo Journal of the Franklin Institute vol 349 no 6pp 1972ndash1988 2012
[12] Y Kao C Wang F Zha and H Cao ldquoStability in mean ofpartial variables for stochastic reactionmdashdiffusion systems withMarkovian switchingrdquo Journal of the Franklin Institute vol 351no 1 pp 500ndash512 2014
[13] N Kumaresan and P Balasubramaniam ldquoOptimal control forstochastic nonlinear singular system using neural networksrdquoComputers amp Mathematics with Applications vol 56 no 9 pp2145ndash2154 2008
[14] L Yu and J Chu ldquoAn LMI approach to guaranteed cost controlof linear uncertain time-delay systemsrdquoAutomatica vol 35 no6 pp 1155ndash1159 1999
[15] S Yin H Luo and S Ding ldquoReal-time implementation of fault-tolerant control systems with performance optimizationrdquo IEEETransactions on Industrial Electronics vol 64 no 5 pp 2402ndash2411 2014
[16] S Yin G Wang and H Karimi ldquoData-driven design of robustfault detection system for wind turbinesrdquoMechatronics 2013
[17] S Yin S X Ding A H A Sari and H Hao ldquoData-drivenmonitoring for stochastic systems and its application on batchprocessrdquo International Journal of Systems Science vol 44 no 7pp 1366ndash1376 2013
[18] S Yin S Ding A Haghani H Hao and P Zhang ldquoA com-parison study of basic datadriven fault diagnosis and process
monitoring methods on the benchmark Tennessee Eastmanprocessrdquo Journal of Process Control vol 22 no 9 pp 1567ndash15812012
[19] Y G Kao and C H Wang ldquoGlobal stability analysis forstochastic coupled reaction-diffusion systems on networksrdquoNonlinear Analysis Real World Applications vol 14 no 3 pp1457ndash1465 2013
[20] E K Boukas ldquoStabilization of stochastic singular nonlinearhybrid systemsrdquo Nonlinear Analysis Theory Methods amp Appli-cations vol 64 no 2 pp 217ndash228 2006
[21] E-K Boukas Control of Singular Systems with Random AbruptChanges Springer Berlin Germany 2008
[22] Y Ding H Zhu S Zhong and Y Zeng ldquoExponential mean-square stability of time-delay singular systems with Markovianswitching and nonlinear perturbationsrdquo Applied Mathematicsand Computation vol 219 no 4 pp 2350ndash2359 2012
[23] S Long S Zhong and Z Liu ldquoStochastic admissibility fora class of singular Markovian jump systems with mode-dependent time delaysrdquoAppliedMathematics and Computationvol 219 no 8 pp 4106ndash4117 2012
[24] Z Wu P Shi H Su and J Chu ldquoAsynchronous 1198682minus 119868infin
filtering for discrete-time stochastic Markov jump systems withrandomly occurred sensor nonlinearitiesrdquo Automatica vol 50no 1 pp 180ndash186 2014
[25] Z Wu P Shi H Su and J Chu ldquoStochastic synchronizationof Markovian jump neural networks with time-varying delayusing sampled-datardquo IEEE Transactions on Cybernetics vol 43no 6 pp 1796ndash1806 2013
[26] S Ma and E-K Boukas ldquoGuaranteed cost control of uncertaindiscrete-time singular Markov jump systems with indefinitequadratic costrdquo International Journal of Robust and NonlinearControl vol 21 no 9 pp 1031ndash1045 2011
[27] G Wang and Q Zhang ldquoRobust control of uncertain singularstochastic systems with Markovian switching via proportional-derivative state feedbackrdquo IET Control Theory amp Applicationsvol 6 no 8 pp 1089ndash1096 2012
[28] J Wang H Wang A Xue and R Lu ldquoDelay-dependent 119867infin
control for singular Markovian jump systems with time delayrdquoNonlinear Analysis Hybrid Systems vol 8 pp 1ndash12 2013
[29] L Wu X Su and P Shi ldquoSliding mode control with bounded1198712gain performance of Markovian jump singular time-delay
systemsrdquo Automatica vol 48 no 8 pp 1929ndash1933 2012[30] Z-GWu J H ParkH Su and J Chu ldquoStochastic stability anal-
ysis for discrete-time singular Markov jump systems with time-varying delay and piecewise-constant transition probabilitiesrdquoJournal of the Franklin Institute vol 349 no 9 pp 2889ndash29022012
[31] J Xiong J Lam H Gao and D W C Ho ldquoOn robust stabi-lization of Markovian jump systems with uncertain switchingprobabilitiesrdquo Automatica vol 41 no 5 pp 897ndash903 2005
[32] J Xiong and J Lam ldquoRobust 1198672control of Markovian jump
systems with uncertain switching probabilitiesrdquo InternationalJournal of Systems Science vol 40 no 3 pp 255ndash265 2009
[33] M Karan P Shi and C Y Kaya ldquoTransition probabilitybounds for the stochastic stability robustness of continuous-and discrete-timeMarkovian jump linear systemsrdquoAutomaticavol 42 no 12 pp 2159ndash2168 2006
[34] L Zhang and E-K Boukas ldquoStability and stabilization ofMarkovian jump linear systems with partly unknown transitionprobabilitiesrdquo Automatica vol 45 no 2 pp 463ndash468 2009
Abstract and Applied Analysis 11
[35] B Du J Lam Y Zou and Z Shu ldquoStability and stabilization forMarkovian jump time-delay systems with partially unknowntransition ratesrdquo IEEE Transactions on Circuits and Systems IRegular Papers vol 60 no 2 pp 341ndash351 2013
[36] L Zhang and J Lam ldquoNecessary and sufficient conditionsfor analysis and synthesis of Markov jump linear systemswith incomplete transition descriptionsrdquo IEEE Transactions onAutomatic Control vol 55 no 7 pp 1695ndash1701 2010
[37] Y Guo and F Zhu ldquoNew results on stability and stabilizationof Markovian jump systems with partly known transitionprobabilitiesrdquoMathematical Problems in Engineering vol 2012Article ID 869842 11 pages 2012
[38] J Lin S Fei and J Shen ldquoDelay-dependent 119867infin
filtering fordiscrete-time singular Markovian jump systems with time-varying delay and partially unknown transition probabilitiesrdquoSignal Processing vol 91 no 2 pp 277ndash289 2011
[39] J Tian Y Li J Zhao and S Zhong ldquoDelay-dependent stochasticstability criteria for Markovian jumping neural networks withmode-dependent time-varying delays and partially knowntransition ratesrdquo Applied Mathematics and Computation vol218 no 9 pp 5769ndash5781 2012
[40] Y Wei J Qiu H R Karimi and M Wang ldquoA new design 119867infin
filtering for contrinuous-time Markovian jump systems withtime-varying delay and partially accessible mode informationrdquoSignal Processing vol 93 pp 2392ndash2407 2013
[41] L Zhang E-K Boukas and J Lam ldquoAnalysis and synthesisof Markov jump linear systems with time-varying delays andpartially known transition probabilitiesrdquo IEEE Transactions onAutomatic Control vol 53 no 10 pp 2458ndash2464 2008
[42] L Zhang and E-K Boukas ldquoMode-dependent 119867infin
filteringfor discrete-time Markovian jump linear systems with partlyunknown transition probabilitiesrdquo Automatica vol 45 no 6pp 1462ndash1467 2009
[43] L Zhang and E-K Boukas ldquoMode-dependent 119867infin
controlfor discrete-time Markovian jump linear systems with partlyunknown transition probabilitiesrdquo International Journal ofRobust and Nonlinear Control vol 19 no 8 pp 868ndash883 2009
[44] Y Zhang Y He M Wu and J Zhang ldquoStabilization forMarkovian jump systems with partial information on transi-tion probability based on free-connection weighting matricesrdquoAutomatica vol 47 no 1 pp 79ndash84 2011
[45] G Zong D Yang L Hou and Q Wang ldquoRobust finite-time119867infincontrol for Markovian jump systems with partially known
transition probabilitiesrdquo Journal of the Franklin Institute vol350 no 6 pp 1562ndash1578 2013
[46] Y Ding H Zhu S Zhong and Y Zhang ldquo1198712minus 119871infin
filteringfor Markovian jump systems with time-varying delays andpartly unknown transition probabilitiesrdquo Communications inNonlinear Science and Numerical Simulation vol 17 no 7 pp3070ndash3081 2012
[47] Q Ma S Xu and Y Zou ldquoStability and synchronizationfor Markovian jump neural networks with partly unknowntransition probabilitiesrdquo Neurocomputing vol 74 pp 3403ndash3411 2011
[48] E Tian D Yue and G Wei ldquoRobust control for Markovianjump systems with partially known transition probabilities andnonlinearitiesrdquo Journal of the Franklin Institute vol 350 no 8pp 2069ndash2083 2013
[49] Y Guo and Z Wang ldquoStability of Markovian jump systemswith generally uncertain transition ratesrdquo Journal of the FranklinInstitute vol 350 no 9 pp 2826ndash2836 2013
[50] L Zhang ldquo119867infinestimation for discrete-time piecewise homoge-
neous Markov jump linear systemsrdquo Automatica vol 45 no 11pp 2570ndash2576 2009
[51] Z Wu H Su and J Chu ldquostate estimation for discreteMarkovian jumping neural networks with time delayrdquo Neuro-computing vol 73 pp 2247ndash2254 2010
[52] I R Petersen and A V Savkin Robust Kalman Filteringfor Signals and Systems with Large Uncertainties BirkhauserBoston Mass USA 1999
[53] B D O Anderson and J B Moore Optimal Filtering Prentice-Hall Upper Saddle River NJ USA 1979
[54] I R Petersen and D C McFarlane ldquoOptimal guaranteedcost control and filtering for uncertain linear systemsrdquo IEEETransactions on Automatic Control vol 39 no 9 pp 1971ndash19771994
[55] S Xu T Chen and J Lam ldquoRobust 119867infin filtering for uncertainMarkovian jump systems with mode-dependent time delaysrdquoIEEE Transactions on Automatic Control vol 48 no 5 pp 900ndash907 2003
[56] J H Kim D Oh and H Park ldquoGuaranteed cost and 119867infin
filtering for time-delay systemsrdquo in Proceeding of the AmericanControl Conference pp 4014ndash4018 IEEE Arlington Va USA2001
[57] L Xie ldquoOutput feedback119867infincontrol of systems with parameter
uncertaintyrdquo International Journal of Control vol 63 no 4 pp741ndash750 1996
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
4 Abstract and Applied Analysis
with generally incomplete transition probability) Given sys-tem (1) with GUTRMatrix (4) can we determine an observer(11) with parameters 119870
1119894and 119870
2119894such that the observer is a
robust guaranteed cost observer for system (1) with GUTRMatrix (4)
Lemma 9 Given any real number 120576 and any matrix Q thematrix inequality 120576(119876 + 119876
119879) le 1205762119879 + 119876119879
minus1119876119879 holds for any
matrix 119879 gt 0
3 Main Results
Theorem 10 Consider the augmented system (13)with GUTRMatrix (4) and the cost function (14) Then the robust guar-anteed cost observer (11) with parameters 119870
1119894and 119870
2119894can
be designed if there exist matrices 119875119894 1198701119894 and 119870
2119894 119894 =
1 2 119904 and symmetric positive definite matrix Q satisfyingthe following LMIs respectively
Case 1 If 119894 notin 119880119894
119896and 119880
119894
119896= 119896119894
1 119896
119894
119898 there exist a set of
symmetric positive definite matrices 119879119894119895isin R119899times119899 (119894 notin 119880
119894
119896 119895 isin
119880119894
119896) such that
119864119879
119891119875119894= 119875119879
119894119864119891ge 0 (15)
[[
[
Π119894+ 119862119879
119891119862119891
119875119894(119860119891119889119894
+ Δ119860119891119889119894
) 1
(119860119891119889119894
+ Δ119860119891119889119894
)119879
119875119894
minus119876 0
lowast lowast 2
]]
]
lt 0 (16)
119875119894minus 119875119895ge 0 forall119895 isin 119880
119894
119906119896 119895 = 119894 (17)
Case 2 If 119894 isin 119880119894
119896 119880119894119896= 119896119894
1 119896
119894
119898 and 119880119894
119906119896= 0 there exist a
set of symmetric positive definite matrices 119881119894119895119897
isin R119899times119899 (119894 119895 isin
119880119894
119896 119897 isin 119880
119894
119906119896) such that
119864119879
119891119875119894= 119875119879
119894119864119891ge 0 (18)
[[
[
Ω119894+ 119862119879
119891119862119891
119875119894(119860119891119889119894
+ Δ119860119891119889119894
) 1
(119860119891119889119894
+ Δ119860119891119889119894
)119879
119875119894
minus119876 0
lowast lowast 2
]]
]
lt 0 (19)
Case 3 If 119894 isin 119880119894
119896and 119880
119894
119906119896= 0 there exist a set of symmetric
positive definite matrices119882119894119895isin R119899times119899 (119894 119895 isin 119880
119894
119896) such that
119864119879
119891119875119894= 119875119879
119894119864119891ge 0 (20)
[[
[
Δ119894+ 119862119879
119891119862119891
119875119894(119860119891119889119894
+ Δ119860119891119889119894
) 1
(119860119891119889119894
+ Δ119860119891119889119894
)119879
119875119894
minus119876 0
lowast lowast 2
]]
]
lt 0 (21)
where
Π119894= (119860119891119894+ Δ119860119891119894)119879
119875119894+ 119875119894(119860119891119894+ Δ119860119891119894)
+ 119876 + sum
119895isin119880119894
119896
119894119895119864119879
119891(119875119895minus 119875119894) + sum
119895isin119880119894
119896
1
41205902
119894119895119879119894119895
Ω119894= (119860119891119894+ Δ119860119891119894)119879
119875119894+ 119875119894(119860119891119894+ Δ119860119891119894)
+ 119876 + sum
119895isin119880119894
119896
119894119895119864119879
119891(119875119895minus 119875119897) + sum
119895isin119880119894
119896
1
41205902
119894119895119881119894119895119897
Δ119894= (119860119891119894+ Δ119860119891119894)119879
119875119894+ 119875119894(119860119891119894+ Δ119860119891119894)
+ 119876 + sum
119895isinS119895 = 119894
119894119895119864119879
119891(119875119895minus 119875119894) + sum
119895isinS119895 = 119894
1
41205902
119894119895119882119894119895
1= [119864119879
119891(119875119894119896119894
1
minus 119875119894) 119864119879
119891(119875119894119896119894
2
minus 119875119894) 119864
119879
119891(119875119894119896119894
119898
minus 119875119894)]
2= diag minus119879
119894119896119894
1
minus119879119894119896119894
119898
1= [119864119879
119891(119875119896119894
1
minus 119875119897) 119864119879
119891(119875119896119894
2
minus 119875119897) 119864
119879
119891(119875119896119894
119898
minus 119875119897)]
2= diag minus119881
119894119896119894
1119897 minus119881
119894119896119894
119898119897
1= [119864119879
119891(1198751minus 119875119894) 119864
119879
119891(119875119894minus1
minus 119875119894)
119864119879
119891(119875119894+1
minus 119875119894) 119864
119879
119891(119875119904minus 119875119894)]
2= diag minus119882
1198941 minus119882
119894119904
(22)
Proof According to Definition 2 and Theorem 1 in [2] wecan derive from (15)ndash(21) that system (13) is regular andimpulse free Let the mode at time 119905 be 119894 and consider thefollowing Lyapunov function with respect to the augmentedsystem (13)
119881(119909119891(119905) 120574 (119905) = 119894) = 119909
119879
119891(119905) 119864119879
119891119875119894119909119891(119905)
+ int
119905
119905minus119889
119909119879
119891(119904) 119876119909
119891(119904) 119889119905
(23)
where 119876 is the symmetric positive definite matrix to be cho-sen and 119875
119894is a matrix satisfying (15)ndash(21)The weak infinites-
imal operatorL of the stochastic process 120574(119905) 119909119891(119905) 119905 ge 0
is presented by
L119881(119909119891(119905) 120574 (119905) = 119894)
= limΔrarr0
1
Δ[119864119891119881 (119909 (119905 + Δ) 120574 (119905 + Δ)) 119909 (119905) 120574 (119905) = 119894
minus119881 (119909 (119905) 120574 (119905) = 119894) ]
= 119909119879
119891(119905) [
[
(119860119891119894+ Δ119860119891119894)119879
119875119894+ 119875119894(119860119891119894+ Δ119860119891119894)
+
119904
sum
119895=1
120587119894119895119864119879
119891119875119895+ 119876]
]
119909119891(119905)
+ 2119909119879
119891(119905) 119875119894(1198601198911119894
+ Δ1198601198911119894
) 119909119891(119905 minus 119889)
minus 119909119879
119891(119905 minus 119889)119876119909
119891(119905 minus 119889)
(24)
Abstract and Applied Analysis 5
Case 1 (119894 notin 119880119894
119896) Note that in this case sum
119895isin119880119894
119906119896119895 = 119894
120587119894119895
=
minussum119895isinU119894119896119895 = 119894
120587119894119895minus 120587119894119894and 120587
119894119895ge 0 119895 isin 119880
119894
119906119896 119895 = 119894 then from
(24) we have
119909119879
119891(119905) [
[
119904
sum
119895=1
120587119894119895119864119879
119891119875119895]
]
119909119891(119905)
= 119909119879
119891(119905) [
[
sum
119895isin119880119894
119896
120587119894119895119864119879
119891119875119895+ sum
119895isin119880119894
119906119896119895 = 119894
120587119894119895119864119879
119891119875119895+ 120587119894119894119864119879
119891119875119895]
]
119909119891(119905)
= 119909119879
119891(119905) [
[
sum
119895isin119880119894
119896
120587119894119895119864119879
119891119875119895+ (minus120587
119894119894minus sum
119895isin119880119894
119896
120587119894119895)119864119879
119891119875119894
+120587119894119894119864119879
119891119875119894]
]
119909119891(119905)
= 119909119879
119891(119905) 119864119879
119891[
[
sum
119895isin119880119894
119896
120587119894119895(119875119895minus 119875119894)]
]
119909119891(119905)
= 119909119879
119891(119905) 119864119879
119891[
[
sum
119895isin119880119894
119896
119894119895(119875119895minus 119875119894) + sum
119895isin119880119894
119896
Δ119894119895(119875119895minus 119875119894)]
]
119909119891(119905)
(25)
On the other hand in view of Lemma 9 we have
sum
119895isin119880119894
119896
Δ119894119895119864119879
119891(119875119895minus 119875119894)
= sum
119895isin119880119894
119896
[1
2Δ119894119895119864119879
119891(119875119895minus 119875119894) +
1
2Δ119894119895119864119879
119891(119875119895minus 119875119894)]
le sum
119895isin119880119894
119896
[(1
2Δ119894119895)
2
119879119894119895+ 119864119879
119891(119875119895minus 119875119894) 119879minus1
119894119895(119875119895minus 119875119894) 119864119891]
le sum
119895isin119880119894
119896
[1
41205902
119894119895119879119894119895+ 119864119879
119891(119875119895minus 119875119894) 119879minus1
119894119895(119875119895minus 119875119894) 119864119891]
(26)
Case 2 (119894 isin 119880119894
119896and119880119894
119906119896= 0) Because of119880119894
119896= 119896119894
1 119896
119894
119898 and
119880119894
119906119896= 119906119894
1 119906
119894
119904minus119898 there must be 119897 isin 119880
119894
119906119896so that 119864119879
119891119875119897ge
119864119879
119891119875119895(for all 119895 isin 119880
119894
119906119896)
119909119879
119891(119905) [
[
119904
sum
119895=1
120587119894119895119864119879
119891119875119895]
]
119909119891(119905)
le 119909119879
119891(119905) [
[
sum
119895isin119880119894
119896
120587119894119895119864119879
119891119875119895minus ( sum
119895isin119880119894
119896119895 = 119894
120587119894119895)119864119879
119891119875119897]
]
119909119891(119905)
= 119909119879
119891(119905) 119864119879
119891[
[
sum
119895isin119880119894
119896
120587119894119895(119875119895minus 119875119897)]
]
119909119891(119905)
= 119909119879
119891(119905) 119864119879
119891[
[
sum
119895isin119880119894
119896
119894119895(119875119895minus 119875119897) + sum
119895isin119880119894
119896
Δ119894119895(119875119895minus 119875119897)]
]
119909119891(119905)
(27)
By using Lemma 9 we have
sum
119895isin119880119894
119896
Δ119894119895119864119879
119891(119875119895minus 119875119897)
= sum
119895isin119880119894
119896
[1
2Δ119894119895119864119879
119891(119875119895minus 119875119897) +
1
2Δ119894119895119864119879
119891(119875119895minus 119875119897)]
le sum
119895isin119880119894
119896
[(1
2Δ119894119895)
2
119881119894119895119897+ 119864119879
119891(119875119895minus 119875119897)119881minus1
119894119895119897(119875119895minus 119875119897) 119864119879
119891]
le sum
119895isin119880119894
119896
[1
41205902
119894119895119881119894119895119897+ 119864119879
119891(119875119895minus 119875119897)119881minus1
119894119895119897(119875119895minus 119875119897) 119864119879
119891]
(28)
Case 3 (119894 isin 119880119894
119896and 119880
119894
119906119896= 0) Consider
119909119879
119891(119905) [
[
119904
sum
119895=1
120587119894119895119864119879
119891119875119895]
]
119909119891(119905)
= 119909119879
119891(119905) 119864119879
119891[
[
119904
sum
119895=1119895 = 119894
120587119894119895(119875119895minus 119875119894)]
]
119909119891(119905)
= 119909119879
119891(119905) 119864119879
119891[
[
119904
sum
119895=1119895 = 119894
119894119895(119875119895minus 119875119894)
+
119904
sum
119895=1119895 = 119894
Δ119894119895(119875119895minus 119875119894)]
]
119909119891(119905)
(29)
Case 1 Substituting (25) and (26) into (24) it results in
L119881 le Λ119879(119905) Φ (119894) Λ (119905) (30)
where Λ119879(119905) = [119909119879
119891(119905) 119909119879
119891(119905 minus 119889)] and
Φ119894=[[
[
Π119894+ 119862119879
119891119862119891
119875119894(119860119891119889119894
+ Δ119860119891119889119894
) 1
(119860119891119889119894
+ Δ119860119891119889119894
)119879
119875119894
minus119876 0
lowast lowast 2
]]
]
(31)
Case 2 Substituting (27) and (28) into (24) it results in
L119881 le Λ119879(119905) Ψ (119894) Λ (119905) (32)
where Λ119879(119905) = [119909119879
119891(119905) 119909119879
119891(119905 minus 119889)] and
Ψ119894=[[
[
Ω119894+ 119862119879
119891119862119891
119875119894(119860119891119889119894
+ Δ119860119891119889119894
) 1
(119860119891119889119894
+ Δ119860119891119889119894
)119879
119875119894
minus119876 0
lowast lowast 2
]]
]
(33)
6 Abstract and Applied Analysis
Case 3 Substituting (29) into (24) we get
L119881 le Λ119879(119905) Γ (119894) Λ (119905) (34)
where Λ119879(119905) = [119909119879
119891(119905) 119909119879
119891(119905 minus 119889)] and
Γ119894=[[
[
Δ119894+ 119862119879
119891119862119891
119875119894(119860119891119889119894
+ Δ119860119891119889119894
) 1
(119860119891119889119894
+ Δ119860119891119889119894
)119879
119875119894
minus119876 0
lowast lowast 2
]]
]
(35)
Similar to [5] usingDynkinrsquos formula we drive for each 119894 isin 119873
lim119879rarrinfin
Eint
119879
0
119909119879
119891(119905) 119909119891(119905) 119889119905 | 120593
119891 1205740= 119894 le 119909
119879
11989101198721199091198910 (36)
By Definition 5 it is easy to see that the augmented system(13) is stochastically stable Furthermore from (16) (19) and(21) we have
L119881 le minus119909119879
119891(119905) 119862119879
119891119862119891119909119891(119905) lt 0 (37)
On the other hand we have
J = Eint
infin
0
119909119879
119891(119905) 119862119879
119891119862119891119909119891(119905) 119889119905 lt minusint
infin
0
L119881119889119905
= minus E lim119905rarrinfin
119881 (119909 (119905) 120574 (119905)) + 119881 (1199090 1205740)
(38)
As the augmented system (13) is stochastically stable itfollows from (38) that 119869 lt 119881(119909
1198910 1199030) From Definition 7 it
is concluded that a robust guaranteed cost for the augmentedsystem (13) can be given by 119869
lowast= 119909119879
1198910(119905)119864119879
1198911199030119875(1199030)1199091198910
+
int0
minus119889119909119879
119891(119905)119876119909
119891(119905)119889119905
In the following based on the above sufficient conditionthe design of robust guaranteed cost observers can be turnedinto the solvability of a system of LMIs
Theorem 11 Consider system (13)with GUTRMatrix (4) andthe cost function (14) If there exist matrices 119884
1119894and 119884
2119894 119894 =
1 2 119904 positive scalars 120576119894 119894 = 1 2 119904 symmetric positive
definite matrix Q and the full rank matrices 1198752119894 and matrices
119875119894= diag(119875
1119894 1198752119894) 119894 = 1 2 119904 satisfying the following
LMIs respectively
Case 1 If 119894 notin 119880119894
119896and 119880
119894
119896= 119896119894
1 119896
119894
119898 a set of positive
definite matrices 119879119894119895isin R119899times119899 (119894 notin 119880
119894
119896 119895 isin 119880
119894
119896) exist such that
119864119879
119891119875119894= 119875119879
119894119864119891ge 0 (39)
[[[[[[[[[
[
1206011119894
1206012119894
1198731
1206013119894
120601119879
2119894minus119876 0 0
119873119879
10 119873
20
120601119879
31198940 0 minus120576
119894119868
]]]]]]]]]
]
lt 0 (40)
119875119894minus 119875119895ge 0 forall119895 isin 119880
119894
119906119896 119895 = 119894 (41)
Case 2 If 119894 isin 119880119894
119896(119880119894
119896= 119896119894
1 119896
119894
119898) and 119880
119894
119906119896= 0 a set of
positive definite matrices119881119894119895119897isin R119899times119899 (119894 119895 isin 119880
119894
119896 119897 isin 119880
119894
119906119896) exist
such that
119864119879
119891119875119894= 119875119879
119894119864119891ge 0 (42)
[[[[[[[[[
[
1205931119894
1205932119894
1198721
1205933119894
120593119879
2119894minus119876 0 0
119872119879
10 119872
20
120593119879
31198940 0 minus120576
119894119868
]]]]]]]]]
]
lt 0 (43)
Case 3 If 119894 isin 119880119894
119896and 119880
119894
119906119896= 0 a set of positive definite
matrices119882119894119895isin R119899times119899 (119894 119895 isin 119880
119894
119896) exist such that
119864119879
119891119894119875119894= 119875119879
119894119864119891119894ge 0 (44)
[[[[[[[[[
[
1205951119894
1205952119894
1198711
1205953119894
120595119879
2119894minus119876 0 0
119871119879
10 1198712
0
120595119879
31198940 0 minus120576
119894119868
]]]]]]]]]
]
lt 0 (45)
where
1206011119894= 1205931119894= 1205951119894
= [1198751119894119860119894+ 119860119879
1198941198751119894
119860119879
1198941198752119894minus 119884119879
1119894minus 119862119879
119894119884119879
2119894
1198752119894119860119894minus 1198841119894minus 1198842119894119862119894
119884119879
1119894+ 1198841119894
]
+ 119876 + 119862119879
119891119862119891+ sum
119895isin119880119894
119896
119894119895119864119879
119891(119875119895minus 119875119894) + sum
119895isin119880119894
119896
1
41205902
119894119895119879119894119895
1206012119894= 1205932119894= 1205952119894= [
1198751119894119860 i 0
1198752119894119860119894minus 1198841119894minus 11988421198941198621198940]
1206013119894= 1205933119894= 1205953119894= [
11987511198941198721119894
11987521198941198721119894minus 11988411198941198721119894minus 11988421198941198722119894
]
1198731= [119864119879
119891(119875119896119894
1
minus 119875119894) 119864119879
119891(119875119896119894
2
minus 119875119894) 119864
119879
119891(119875119896119894
119898
minus 119875119894)]
1198732= diag minus119879
119894119896119894
1
minus119879119894119896119894
119898
1198721= [119864119879
119891(119875119896119894
1
minus 119875119894) 119864119879
119891(119875119896119894
2
minus 119875119894) 119864
119879
119891(119875119896119894
119898
minus 119875119894)]
1198722= diag minus119881
119894119896119894
1119897 minus119881
119894119896119894
119898119897
1198711= [119864119879
119891(1198751minus 119875119894) 119864
119879
119891(119875119894minus1
minus 119875119894)
119864119879
119891(119875119894+1
minus 119875119894) 119864
119879
119891(119875119904minus 119875119894)]
1198712= diag minus119882
1198941 minus119882
119894119904
(46)
Abstract and Applied Analysis 7
Then a suitable robust guaranteed cost observer in theform of (11) has parameters as follows
1198701119894= 119875minus1
11198941198841119894 119870
2119894= 119875minus1
21198941198842119894
(47)
and 119869lowast is a robust guaranteed cost for system (13) with GUTRMatrix (4)
Proof Define
1198601
119894=
[[[[
[
119860119879
119891119894119875119894+ 119875119894119860119891119894+ 119876 + sum
119895isin119880119894
119896
119894119895(119875119895minus 119875119894) + sum
119895isin119880119894
119896
1
41205902
119894119895119879119894119895+ 119862119879
119891119862119891
119875119894119860119891119889119894
1198731
119860119879
119891119889119894119875119894
minus119876 0
lowast lowast 1198732
]]]]
]
(48)
1198602
119894=
[[[[
[
119860119879
119891119894119875119894+ 119875119894119860119891119894+ 119876 + sum
119895isin119880119894
119896
119894119895(119875119895minus 119875119894) + sum
119895isin119880119894
119896
1
41205902
119894119895119881119894119895119897+ 119862119879
119891119862119891
119875119894119860119891119889119894
1198721
119860119879
119891119889119894119875119894
minus119876 0
lowast lowast 1198722
]]]]
]
(49)
1198603
119894=
[[[[
[
119860119879
119891119894119875119894+ 119875119894119860119891119894+ 119876 + sum
119895isin119880119894
119896
119894119895(119875119895minus 119875119894) + sum
119895isin119880119894
119896
1
41205902
119894119895119882119894119895+ 119862119879
119891119862119891
119875119894119860119891119889119894
1198711
119860119879
119891119889119894119875119894
minus119876 0
lowast lowast 1198712
]]]]
]
lt 0 (50)
Then (16) is equivalent to
1198601
119894+ [
[
119875119894119872119891119894
0
0
]
]
119865119894[119873119891119894 1198731198911119894 0]
+ [119873119891119894 1198651198911119894 0]119879
119865119879
119894[
[
119875119894119872119891119894
0
0
]
]
119879
lt 0
(51)
Then (19) is equivalent to
1198602
119894+ [
[
119875119894119872119891119894
0
0
]
]
119865119894[119873119891119894 1198731198911119894 0]
+ [119873119891119894 1198651198911119894 0]119879
119865119879
119894[
[
119875119894119872119891119894
0
0
]
]
119879
lt 0
(52)
Then (21) is equivalent to
1198603
119894+ [
[
119875119894119872119891119894
0
0
]
]
119865119894[119873119891119894 1198731198911119894 0]
+ [119873119891119894 1198651198911119894 0]119879
119865119879
119894[
[
119875119894119872119891119894
0
0
]
]
119879
lt 0
(53)
By applying Lemma 24 in [57] (50) (51) and (52) hold forall uncertainties119865i satisfying119865
119879
119894119865119894lt 119868 if and only if there exist
positive scalars 120576119894 119894 = 1 2 119904 such that
1198601
119894+ 120576minus1
119894[
[
119875119894119872119891119894
0
0
]
]
[
[
119875119894119872119891119894
0
0
]
]
119879
+ 120576119894[119873119891119894 1198651198911119894 0]
119879
[119873119891119894 1198651198911119894 0] lt 0
1198602
119894+ 120576minus1
119894[
[
119875119894119872119891119894
0
0
]
]
[
[
119875119894119872119891119894
0
0
]
]
119879
+ 120576119894[119873119891119894 1198651198911119894 0]
119879
[119873119891119894 1198651198911119894 0] lt 0
1198603
119894+ 120576minus1
119894[
[
119875119894119872119891119894
0
0
]
]
[
[
119875119894119872119891119894
0
0
]
]
119879
+ 120576119894[119873119891119894 1198651198911119894 0]
119879
[119873119891119894 1198651198911119894 0] lt 0
(54)
Let 119875119894= diag(119875
1119894 1198752119894) and using (47) we can conclude from
Schur complement results that the above matrix inequalitiesare equivalent to the coupled LMIs (40) (43) and (45)It further follows from Theorem 10 that 119869
lowast is a robustguaranteed cost for system (13) with (4)
Remark 12 The solution of LMIs (39)ndash(45) parameterizesthe set of the proposed robust guaranteed cost observersThis parameterized representation can be used to design theguaranteed cost observer with some additional performanceconstraints By applying the methods in [14] the suboptimal
8 Abstract and Applied Analysis
guaranteed cost observer can be determined by solving acertain optimization problem This is the following theorem
Theorem 13 Consider system (13)with GUTRMatrix (4) andthe cost function (14) and suppose that the initial conditions 119903
0
and 1199091198910
are known if the following optimization problem
min1198761198751119894 1198752119894 120576119894 1198841119894 and 1198842119894
119869lowast
st LMIs (39) ndash (45)(55)
has a solution 119876 1198751119894 1198752119894 120576119894 1198841119894 and 119884
2119894 119894 = 1 2 119904
then the observer (11) is a suboptimal guaranteed costobserver for system (1) where 119869
lowast= 119909119879
1198910119864119879
1198911199030119875(1199030)1199091198910
+
tr(int0minus119889
1199091198910(119905)1199091198910(119905)119909119879
1198910119889119905119876)
Proof It follows from Theorem 11 that the observer (11)constructed in terms of the solution 119876 119875
1119894 1198752119894 120576119894 1198841119894 and
1198842119894 119894 = 1 2 119904 is a robust guaranteed cost observer By
noting that
int
0
minus119889
119909119879
1198910(119905) 119876119909
1198910(119905) 119889119905 = int
0
minus119889
tr (1199091198791198910
(119905) 1198761199091198910
(119905)) 119889119905
= tr(int0
minus119889
119909119879
1198910(119905) 1199091198910
(119905) 119889119905119876)
(56)
it follows that the suboptimal guaranteed cost observerproblem is turned into the minimization problem (55)
Remark 14 Theorem 13 gives the suboptimal guaranteed costobserver conditions of a class of linear Markovian jump-ing time-delay systems with generally incomplete transitionprobability and LMI constraints which can be easily solvedby the LMI toolbox in MATLAB
4 Numerical Example
In this section a numerical example is presented todemonstrate the effectiveness of the method mentioned inTheorem 11 Consider a 2-dimensional system (1) with 3Markovian switching modes In this numerical example thesingular system matrix is set as 119864 = [ 1 0
0 0] and the 3-mode
transition rate matrix is Λ = [minus32
2
15 21 minus36
] where Δ11 Δ31
isin
[minus015 015] Δ23 Δ33
isin [minus012 012] and Δ32
isin [minus01 01]The other system matrices are as follows
For mode 119894 = 1 there are
1198601= [
minus32 065
1 02] 119860
1198891= [
02 05
1 minus068]
1198621= [
[
12 065
minus65 19
minus021 minus18
]
]
1198621198891
= [
[
minus36 minus105
21 096
021 minus086
]
]
11987211
= [minus02
08] 119872
21= [
[
025
0875
minus2
]
]
11987311
= [minus12 31] 11987321
= [minus069 minus42]
(57)
For mode 119894 = 2 there are
1198602= [
minus1 6
2 minus36] 119860
1198892= [
minus31 minus16
3 075]
1198622= [
[
9 minus25
035 minus2
36 minus18
]
]
1198621198892
= [
[
089 minus6
minus12 09
minus24 6
]
]
11987212
= [23
minus4] 119872
22= [
[
075
minus36
25
]
]
11987312
= [minus72 minus6] 11987322
= [1 2]
(58)
For mode 119894 = 3 there are
1198603= [
minus106 29
minus03 36] 119860
1198893= [
minus56 minus12
minus3 45]
1198623= [
[
minus3 minus036
015 minus18
09 minus5
]
]
1198621198893
= [
[
minus165 5
minus12 265
minus098 minus56
]
]
11987213
= [minus82
minus03] 119872
23= [
[
minus052
25
minus36
]
]
11987313
= [105 minus5] 11987323
= [minus72 minus126]
(59)
Then we set the error state matrix 119871 = [ minus45 062 minus6
] and thepositive scalars in Theorem 11 are 120576
1= 02 120576
2= 015 120576
3=
032 According to the definitions of augmented statematricesin (12) we can easily obtain the following parameter matricesin Theorem 11 by MATLAB
11988411
= [minus84521006 00127
00127 84509001]
11988421
= [002 01291 01435
minus23520 minus04080 minus08106]
11988412
= [minus170991 269626
269626 minus246750]
11988422
= [minus200744 minus132941 529388
216893 180120 minus506693]
11988413
= [minus6751329 224456
224456 minus8976976]
11988423
= [minus136021 minus1461726 540500
minus34324 minus195125 minus103068]
1198751=[[[
[
63029 0 0 0
0 48620 0 0
0 0 22914 0
0 0 0 03169
]]]
]
Abstract and Applied Analysis 9
1198752=[[[
[
08914 0 0 0
0 12505 0 0
0 0 73629 0
0 0 0 30056
]]]
]
1198753=[[[
[
30265 0 0 0
0 02156 0 0
0 0 08965 0
0 0 0 10002
]]]
]
119876 =[[[
[
05000 0 0 0
0 05001 0 0
0 0 05001 0
0 0 0 05001
]]]
]
11987911
=[[[
[
34173214 minus8707765 0 0
minus8707765 4167216 0 0
0 0 22263598 minus3207456
0 0 minus3207456 12263101
]]]
]
11987923
=[[[
[
37753231 minus27999330 0 0
minus27999330 28107685 0 0
0 0 106907366 minus107432750
0 0 minus107432750 108555053
]]]
]
11987931
=[[[
[
9518504 minus5399245 0 0
minus5399245 8962029 0 0
0 0 14773012 minus2077540
0 0 minus2077540 14791256
]]]
]
11987932
=[[[
[
21617695 minus12094164 0 0
minus12094164 20371205 0 0
0 0 14786 minus09283
0 0 minus09283 14794
]]]
]
11987933
=[[[
[
14939313 minus8398780 0 0
minus8398780 14073689 0 0
0 0 1478123 minus1336452
0 0 minus1336452 2459347
]]]
]
11988111
=[[[
[
16650 0 0 0
0 16650 0 0
0 0 16650 0
0 0 0 16650
]]]
]
11988231
=[[[
[
15426 0 0 0
0 16650 0 0
0 0 16662 0
0 0 0 16650
]]]
]
11988232
=[[[
[
15428 0 0 0
0 16650 0 0
0 0 16622 0
0 0 0 16650
]]]
]
(60)
Therefore we can design a linear memoryless observer as(11) with the constant matrices
11987011
= 119875minus1
1111988411
= [minus13409860 00020
00026 17381530]
11987021
= 119875minus1
2111988421
= [00087 00563 00626
minus74219 minus12875 minus25579]
11987012
= 119875minus1
1211988412
= [minus191823 302475
215615 minus197321]
11987022
= 119875minus1
2211988422
= [minus27264 minus18056 71899
72163 59928 minus168583]
11987013
= 119875minus1
1311988413
= [minus2231402 74186
1041076 minus41637180]
11987023
= 119875minus1
2311988423
= [minus151724 minus1630481 602900
minus34317 minus195086 minus103047]
(61)
Finally the observer (11) with the above parameter matri-ces for this numerical example is a suboptimal guaranteedcost observer byTheorems 11 and 13
5 Conclusions
In this paper the robust guaranteed cost observer problemfor a class of uncertain descriptor time-delay systems withMarkovian jumping parameters and generally uncertaintransition rates is studied by using LMI method In thisGUTR singular model each transition rate can be com-pletely unknown or only its estimate value is known Theparameterrsquos uncertainty is time varying and is assumed tobe norm-bounded Memoryless guaranteed cost observersare designed in terms of a set of linear coupled matrixinequalities The suboptimal guaranteed cost observer isdesigned by solving a certain optimization problem Ourresults can be applicable to the general Markovian jumpsystems with bounded uncertain or partly uncertain TRmatrix
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This research is supported by the National NaturalScience Foundations of China (10701020 1097102260973048 61272077 60974025 60673101 and 60939003)NCET-08-0755 National 863 Plan Project (2008AA04Z401 2009AA043404) SDPW IMZQWH010016the Natural Science Foundation of Shandong Province (noZR2012FM006) the Scientific and Technological Projectof Shandong Province (no 2007GG3WZ04016) and theNatural Science Foundation of Guangxi Autonomous Region(no 2012GXNSFBA053003)
10 Abstract and Applied Analysis
References
[1] L Dai Singular Control Systems Springer Berlin Germany1989
[2] S Xu P V Dooren R Stefan and J Lam ldquoRobust stability andstabilization for singular systemswith state delay and parameteruncertaintyrdquo IEEE Transactions on Automatic Control vol 47no 7 pp 1122ndash1128 2002
[3] M Fu and G Duan ldquoRobust guaranteed cost observer foruncertain descriptor time-delay systems withMarkovian jump-ing parametersrdquo Acta Automatica Sinica vol 31 pp 479ndash4832005
[4] Y Q Xia E-K Boukas P Shi and J H Zhang ldquoStabilityand stabilization of continuous-time singular hybrid systemsrdquoAutomatica vol 45 no 6 pp 1504ndash1509 2009
[5] S Xu and J Lam Robust Control and Filtering of SingularSystems Springer Berlin Germany 2006
[6] Y-Y Cao and J Lam ldquoRobust 119867infin
control of uncertainMarkovian jump systems with time-delayrdquo IEEE Transactionson Automatic Control vol 45 no 1 pp 77ndash83 2000
[7] Y Li and Y Kao ldquoStability of stochastic reaction-diffusion sys-tems with Markovian switching and impulsive perturbationsrdquoMathematical Problems in Engineering vol 2012 Article ID429568 13 pages 2012
[8] X Mao and C Yuan Stochastic Differential Equations withMarkovian Switching Imperial College Press London UK2006
[9] E-K Boukas Stochastic Switching Systems Analysis and DesignBirkhauser Basel Switzerland 2005
[10] Y Kao C Wang and L Zhang ldquoDelay-dependent exponentialstability of impulsive markovian jumping Cohen-Grossbergneural networks with reaction-diffusion and mixed delaysrdquoNeural Processing Letters vol 38 no 3 pp 321ndash346 2013
[11] Y-G Kao J-F Guo C-H Wang and X-Q Sun ldquoDelay-dependent robust exponential stability of Markovian jump-ing reaction-diffusion Cohen-Grossberg neural networks withmixed delaysrdquo Journal of the Franklin Institute vol 349 no 6pp 1972ndash1988 2012
[12] Y Kao C Wang F Zha and H Cao ldquoStability in mean ofpartial variables for stochastic reactionmdashdiffusion systems withMarkovian switchingrdquo Journal of the Franklin Institute vol 351no 1 pp 500ndash512 2014
[13] N Kumaresan and P Balasubramaniam ldquoOptimal control forstochastic nonlinear singular system using neural networksrdquoComputers amp Mathematics with Applications vol 56 no 9 pp2145ndash2154 2008
[14] L Yu and J Chu ldquoAn LMI approach to guaranteed cost controlof linear uncertain time-delay systemsrdquoAutomatica vol 35 no6 pp 1155ndash1159 1999
[15] S Yin H Luo and S Ding ldquoReal-time implementation of fault-tolerant control systems with performance optimizationrdquo IEEETransactions on Industrial Electronics vol 64 no 5 pp 2402ndash2411 2014
[16] S Yin G Wang and H Karimi ldquoData-driven design of robustfault detection system for wind turbinesrdquoMechatronics 2013
[17] S Yin S X Ding A H A Sari and H Hao ldquoData-drivenmonitoring for stochastic systems and its application on batchprocessrdquo International Journal of Systems Science vol 44 no 7pp 1366ndash1376 2013
[18] S Yin S Ding A Haghani H Hao and P Zhang ldquoA com-parison study of basic datadriven fault diagnosis and process
monitoring methods on the benchmark Tennessee Eastmanprocessrdquo Journal of Process Control vol 22 no 9 pp 1567ndash15812012
[19] Y G Kao and C H Wang ldquoGlobal stability analysis forstochastic coupled reaction-diffusion systems on networksrdquoNonlinear Analysis Real World Applications vol 14 no 3 pp1457ndash1465 2013
[20] E K Boukas ldquoStabilization of stochastic singular nonlinearhybrid systemsrdquo Nonlinear Analysis Theory Methods amp Appli-cations vol 64 no 2 pp 217ndash228 2006
[21] E-K Boukas Control of Singular Systems with Random AbruptChanges Springer Berlin Germany 2008
[22] Y Ding H Zhu S Zhong and Y Zeng ldquoExponential mean-square stability of time-delay singular systems with Markovianswitching and nonlinear perturbationsrdquo Applied Mathematicsand Computation vol 219 no 4 pp 2350ndash2359 2012
[23] S Long S Zhong and Z Liu ldquoStochastic admissibility fora class of singular Markovian jump systems with mode-dependent time delaysrdquoAppliedMathematics and Computationvol 219 no 8 pp 4106ndash4117 2012
[24] Z Wu P Shi H Su and J Chu ldquoAsynchronous 1198682minus 119868infin
filtering for discrete-time stochastic Markov jump systems withrandomly occurred sensor nonlinearitiesrdquo Automatica vol 50no 1 pp 180ndash186 2014
[25] Z Wu P Shi H Su and J Chu ldquoStochastic synchronizationof Markovian jump neural networks with time-varying delayusing sampled-datardquo IEEE Transactions on Cybernetics vol 43no 6 pp 1796ndash1806 2013
[26] S Ma and E-K Boukas ldquoGuaranteed cost control of uncertaindiscrete-time singular Markov jump systems with indefinitequadratic costrdquo International Journal of Robust and NonlinearControl vol 21 no 9 pp 1031ndash1045 2011
[27] G Wang and Q Zhang ldquoRobust control of uncertain singularstochastic systems with Markovian switching via proportional-derivative state feedbackrdquo IET Control Theory amp Applicationsvol 6 no 8 pp 1089ndash1096 2012
[28] J Wang H Wang A Xue and R Lu ldquoDelay-dependent 119867infin
control for singular Markovian jump systems with time delayrdquoNonlinear Analysis Hybrid Systems vol 8 pp 1ndash12 2013
[29] L Wu X Su and P Shi ldquoSliding mode control with bounded1198712gain performance of Markovian jump singular time-delay
systemsrdquo Automatica vol 48 no 8 pp 1929ndash1933 2012[30] Z-GWu J H ParkH Su and J Chu ldquoStochastic stability anal-
ysis for discrete-time singular Markov jump systems with time-varying delay and piecewise-constant transition probabilitiesrdquoJournal of the Franklin Institute vol 349 no 9 pp 2889ndash29022012
[31] J Xiong J Lam H Gao and D W C Ho ldquoOn robust stabi-lization of Markovian jump systems with uncertain switchingprobabilitiesrdquo Automatica vol 41 no 5 pp 897ndash903 2005
[32] J Xiong and J Lam ldquoRobust 1198672control of Markovian jump
systems with uncertain switching probabilitiesrdquo InternationalJournal of Systems Science vol 40 no 3 pp 255ndash265 2009
[33] M Karan P Shi and C Y Kaya ldquoTransition probabilitybounds for the stochastic stability robustness of continuous-and discrete-timeMarkovian jump linear systemsrdquoAutomaticavol 42 no 12 pp 2159ndash2168 2006
[34] L Zhang and E-K Boukas ldquoStability and stabilization ofMarkovian jump linear systems with partly unknown transitionprobabilitiesrdquo Automatica vol 45 no 2 pp 463ndash468 2009
Abstract and Applied Analysis 11
[35] B Du J Lam Y Zou and Z Shu ldquoStability and stabilization forMarkovian jump time-delay systems with partially unknowntransition ratesrdquo IEEE Transactions on Circuits and Systems IRegular Papers vol 60 no 2 pp 341ndash351 2013
[36] L Zhang and J Lam ldquoNecessary and sufficient conditionsfor analysis and synthesis of Markov jump linear systemswith incomplete transition descriptionsrdquo IEEE Transactions onAutomatic Control vol 55 no 7 pp 1695ndash1701 2010
[37] Y Guo and F Zhu ldquoNew results on stability and stabilizationof Markovian jump systems with partly known transitionprobabilitiesrdquoMathematical Problems in Engineering vol 2012Article ID 869842 11 pages 2012
[38] J Lin S Fei and J Shen ldquoDelay-dependent 119867infin
filtering fordiscrete-time singular Markovian jump systems with time-varying delay and partially unknown transition probabilitiesrdquoSignal Processing vol 91 no 2 pp 277ndash289 2011
[39] J Tian Y Li J Zhao and S Zhong ldquoDelay-dependent stochasticstability criteria for Markovian jumping neural networks withmode-dependent time-varying delays and partially knowntransition ratesrdquo Applied Mathematics and Computation vol218 no 9 pp 5769ndash5781 2012
[40] Y Wei J Qiu H R Karimi and M Wang ldquoA new design 119867infin
filtering for contrinuous-time Markovian jump systems withtime-varying delay and partially accessible mode informationrdquoSignal Processing vol 93 pp 2392ndash2407 2013
[41] L Zhang E-K Boukas and J Lam ldquoAnalysis and synthesisof Markov jump linear systems with time-varying delays andpartially known transition probabilitiesrdquo IEEE Transactions onAutomatic Control vol 53 no 10 pp 2458ndash2464 2008
[42] L Zhang and E-K Boukas ldquoMode-dependent 119867infin
filteringfor discrete-time Markovian jump linear systems with partlyunknown transition probabilitiesrdquo Automatica vol 45 no 6pp 1462ndash1467 2009
[43] L Zhang and E-K Boukas ldquoMode-dependent 119867infin
controlfor discrete-time Markovian jump linear systems with partlyunknown transition probabilitiesrdquo International Journal ofRobust and Nonlinear Control vol 19 no 8 pp 868ndash883 2009
[44] Y Zhang Y He M Wu and J Zhang ldquoStabilization forMarkovian jump systems with partial information on transi-tion probability based on free-connection weighting matricesrdquoAutomatica vol 47 no 1 pp 79ndash84 2011
[45] G Zong D Yang L Hou and Q Wang ldquoRobust finite-time119867infincontrol for Markovian jump systems with partially known
transition probabilitiesrdquo Journal of the Franklin Institute vol350 no 6 pp 1562ndash1578 2013
[46] Y Ding H Zhu S Zhong and Y Zhang ldquo1198712minus 119871infin
filteringfor Markovian jump systems with time-varying delays andpartly unknown transition probabilitiesrdquo Communications inNonlinear Science and Numerical Simulation vol 17 no 7 pp3070ndash3081 2012
[47] Q Ma S Xu and Y Zou ldquoStability and synchronizationfor Markovian jump neural networks with partly unknowntransition probabilitiesrdquo Neurocomputing vol 74 pp 3403ndash3411 2011
[48] E Tian D Yue and G Wei ldquoRobust control for Markovianjump systems with partially known transition probabilities andnonlinearitiesrdquo Journal of the Franklin Institute vol 350 no 8pp 2069ndash2083 2013
[49] Y Guo and Z Wang ldquoStability of Markovian jump systemswith generally uncertain transition ratesrdquo Journal of the FranklinInstitute vol 350 no 9 pp 2826ndash2836 2013
[50] L Zhang ldquo119867infinestimation for discrete-time piecewise homoge-
neous Markov jump linear systemsrdquo Automatica vol 45 no 11pp 2570ndash2576 2009
[51] Z Wu H Su and J Chu ldquostate estimation for discreteMarkovian jumping neural networks with time delayrdquo Neuro-computing vol 73 pp 2247ndash2254 2010
[52] I R Petersen and A V Savkin Robust Kalman Filteringfor Signals and Systems with Large Uncertainties BirkhauserBoston Mass USA 1999
[53] B D O Anderson and J B Moore Optimal Filtering Prentice-Hall Upper Saddle River NJ USA 1979
[54] I R Petersen and D C McFarlane ldquoOptimal guaranteedcost control and filtering for uncertain linear systemsrdquo IEEETransactions on Automatic Control vol 39 no 9 pp 1971ndash19771994
[55] S Xu T Chen and J Lam ldquoRobust 119867infin filtering for uncertainMarkovian jump systems with mode-dependent time delaysrdquoIEEE Transactions on Automatic Control vol 48 no 5 pp 900ndash907 2003
[56] J H Kim D Oh and H Park ldquoGuaranteed cost and 119867infin
filtering for time-delay systemsrdquo in Proceeding of the AmericanControl Conference pp 4014ndash4018 IEEE Arlington Va USA2001
[57] L Xie ldquoOutput feedback119867infincontrol of systems with parameter
uncertaintyrdquo International Journal of Control vol 63 no 4 pp741ndash750 1996
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Abstract and Applied Analysis 5
Case 1 (119894 notin 119880119894
119896) Note that in this case sum
119895isin119880119894
119906119896119895 = 119894
120587119894119895
=
minussum119895isinU119894119896119895 = 119894
120587119894119895minus 120587119894119894and 120587
119894119895ge 0 119895 isin 119880
119894
119906119896 119895 = 119894 then from
(24) we have
119909119879
119891(119905) [
[
119904
sum
119895=1
120587119894119895119864119879
119891119875119895]
]
119909119891(119905)
= 119909119879
119891(119905) [
[
sum
119895isin119880119894
119896
120587119894119895119864119879
119891119875119895+ sum
119895isin119880119894
119906119896119895 = 119894
120587119894119895119864119879
119891119875119895+ 120587119894119894119864119879
119891119875119895]
]
119909119891(119905)
= 119909119879
119891(119905) [
[
sum
119895isin119880119894
119896
120587119894119895119864119879
119891119875119895+ (minus120587
119894119894minus sum
119895isin119880119894
119896
120587119894119895)119864119879
119891119875119894
+120587119894119894119864119879
119891119875119894]
]
119909119891(119905)
= 119909119879
119891(119905) 119864119879
119891[
[
sum
119895isin119880119894
119896
120587119894119895(119875119895minus 119875119894)]
]
119909119891(119905)
= 119909119879
119891(119905) 119864119879
119891[
[
sum
119895isin119880119894
119896
119894119895(119875119895minus 119875119894) + sum
119895isin119880119894
119896
Δ119894119895(119875119895minus 119875119894)]
]
119909119891(119905)
(25)
On the other hand in view of Lemma 9 we have
sum
119895isin119880119894
119896
Δ119894119895119864119879
119891(119875119895minus 119875119894)
= sum
119895isin119880119894
119896
[1
2Δ119894119895119864119879
119891(119875119895minus 119875119894) +
1
2Δ119894119895119864119879
119891(119875119895minus 119875119894)]
le sum
119895isin119880119894
119896
[(1
2Δ119894119895)
2
119879119894119895+ 119864119879
119891(119875119895minus 119875119894) 119879minus1
119894119895(119875119895minus 119875119894) 119864119891]
le sum
119895isin119880119894
119896
[1
41205902
119894119895119879119894119895+ 119864119879
119891(119875119895minus 119875119894) 119879minus1
119894119895(119875119895minus 119875119894) 119864119891]
(26)
Case 2 (119894 isin 119880119894
119896and119880119894
119906119896= 0) Because of119880119894
119896= 119896119894
1 119896
119894
119898 and
119880119894
119906119896= 119906119894
1 119906
119894
119904minus119898 there must be 119897 isin 119880
119894
119906119896so that 119864119879
119891119875119897ge
119864119879
119891119875119895(for all 119895 isin 119880
119894
119906119896)
119909119879
119891(119905) [
[
119904
sum
119895=1
120587119894119895119864119879
119891119875119895]
]
119909119891(119905)
le 119909119879
119891(119905) [
[
sum
119895isin119880119894
119896
120587119894119895119864119879
119891119875119895minus ( sum
119895isin119880119894
119896119895 = 119894
120587119894119895)119864119879
119891119875119897]
]
119909119891(119905)
= 119909119879
119891(119905) 119864119879
119891[
[
sum
119895isin119880119894
119896
120587119894119895(119875119895minus 119875119897)]
]
119909119891(119905)
= 119909119879
119891(119905) 119864119879
119891[
[
sum
119895isin119880119894
119896
119894119895(119875119895minus 119875119897) + sum
119895isin119880119894
119896
Δ119894119895(119875119895minus 119875119897)]
]
119909119891(119905)
(27)
By using Lemma 9 we have
sum
119895isin119880119894
119896
Δ119894119895119864119879
119891(119875119895minus 119875119897)
= sum
119895isin119880119894
119896
[1
2Δ119894119895119864119879
119891(119875119895minus 119875119897) +
1
2Δ119894119895119864119879
119891(119875119895minus 119875119897)]
le sum
119895isin119880119894
119896
[(1
2Δ119894119895)
2
119881119894119895119897+ 119864119879
119891(119875119895minus 119875119897)119881minus1
119894119895119897(119875119895minus 119875119897) 119864119879
119891]
le sum
119895isin119880119894
119896
[1
41205902
119894119895119881119894119895119897+ 119864119879
119891(119875119895minus 119875119897)119881minus1
119894119895119897(119875119895minus 119875119897) 119864119879
119891]
(28)
Case 3 (119894 isin 119880119894
119896and 119880
119894
119906119896= 0) Consider
119909119879
119891(119905) [
[
119904
sum
119895=1
120587119894119895119864119879
119891119875119895]
]
119909119891(119905)
= 119909119879
119891(119905) 119864119879
119891[
[
119904
sum
119895=1119895 = 119894
120587119894119895(119875119895minus 119875119894)]
]
119909119891(119905)
= 119909119879
119891(119905) 119864119879
119891[
[
119904
sum
119895=1119895 = 119894
119894119895(119875119895minus 119875119894)
+
119904
sum
119895=1119895 = 119894
Δ119894119895(119875119895minus 119875119894)]
]
119909119891(119905)
(29)
Case 1 Substituting (25) and (26) into (24) it results in
L119881 le Λ119879(119905) Φ (119894) Λ (119905) (30)
where Λ119879(119905) = [119909119879
119891(119905) 119909119879
119891(119905 minus 119889)] and
Φ119894=[[
[
Π119894+ 119862119879
119891119862119891
119875119894(119860119891119889119894
+ Δ119860119891119889119894
) 1
(119860119891119889119894
+ Δ119860119891119889119894
)119879
119875119894
minus119876 0
lowast lowast 2
]]
]
(31)
Case 2 Substituting (27) and (28) into (24) it results in
L119881 le Λ119879(119905) Ψ (119894) Λ (119905) (32)
where Λ119879(119905) = [119909119879
119891(119905) 119909119879
119891(119905 minus 119889)] and
Ψ119894=[[
[
Ω119894+ 119862119879
119891119862119891
119875119894(119860119891119889119894
+ Δ119860119891119889119894
) 1
(119860119891119889119894
+ Δ119860119891119889119894
)119879
119875119894
minus119876 0
lowast lowast 2
]]
]
(33)
6 Abstract and Applied Analysis
Case 3 Substituting (29) into (24) we get
L119881 le Λ119879(119905) Γ (119894) Λ (119905) (34)
where Λ119879(119905) = [119909119879
119891(119905) 119909119879
119891(119905 minus 119889)] and
Γ119894=[[
[
Δ119894+ 119862119879
119891119862119891
119875119894(119860119891119889119894
+ Δ119860119891119889119894
) 1
(119860119891119889119894
+ Δ119860119891119889119894
)119879
119875119894
minus119876 0
lowast lowast 2
]]
]
(35)
Similar to [5] usingDynkinrsquos formula we drive for each 119894 isin 119873
lim119879rarrinfin
Eint
119879
0
119909119879
119891(119905) 119909119891(119905) 119889119905 | 120593
119891 1205740= 119894 le 119909
119879
11989101198721199091198910 (36)
By Definition 5 it is easy to see that the augmented system(13) is stochastically stable Furthermore from (16) (19) and(21) we have
L119881 le minus119909119879
119891(119905) 119862119879
119891119862119891119909119891(119905) lt 0 (37)
On the other hand we have
J = Eint
infin
0
119909119879
119891(119905) 119862119879
119891119862119891119909119891(119905) 119889119905 lt minusint
infin
0
L119881119889119905
= minus E lim119905rarrinfin
119881 (119909 (119905) 120574 (119905)) + 119881 (1199090 1205740)
(38)
As the augmented system (13) is stochastically stable itfollows from (38) that 119869 lt 119881(119909
1198910 1199030) From Definition 7 it
is concluded that a robust guaranteed cost for the augmentedsystem (13) can be given by 119869
lowast= 119909119879
1198910(119905)119864119879
1198911199030119875(1199030)1199091198910
+
int0
minus119889119909119879
119891(119905)119876119909
119891(119905)119889119905
In the following based on the above sufficient conditionthe design of robust guaranteed cost observers can be turnedinto the solvability of a system of LMIs
Theorem 11 Consider system (13)with GUTRMatrix (4) andthe cost function (14) If there exist matrices 119884
1119894and 119884
2119894 119894 =
1 2 119904 positive scalars 120576119894 119894 = 1 2 119904 symmetric positive
definite matrix Q and the full rank matrices 1198752119894 and matrices
119875119894= diag(119875
1119894 1198752119894) 119894 = 1 2 119904 satisfying the following
LMIs respectively
Case 1 If 119894 notin 119880119894
119896and 119880
119894
119896= 119896119894
1 119896
119894
119898 a set of positive
definite matrices 119879119894119895isin R119899times119899 (119894 notin 119880
119894
119896 119895 isin 119880
119894
119896) exist such that
119864119879
119891119875119894= 119875119879
119894119864119891ge 0 (39)
[[[[[[[[[
[
1206011119894
1206012119894
1198731
1206013119894
120601119879
2119894minus119876 0 0
119873119879
10 119873
20
120601119879
31198940 0 minus120576
119894119868
]]]]]]]]]
]
lt 0 (40)
119875119894minus 119875119895ge 0 forall119895 isin 119880
119894
119906119896 119895 = 119894 (41)
Case 2 If 119894 isin 119880119894
119896(119880119894
119896= 119896119894
1 119896
119894
119898) and 119880
119894
119906119896= 0 a set of
positive definite matrices119881119894119895119897isin R119899times119899 (119894 119895 isin 119880
119894
119896 119897 isin 119880
119894
119906119896) exist
such that
119864119879
119891119875119894= 119875119879
119894119864119891ge 0 (42)
[[[[[[[[[
[
1205931119894
1205932119894
1198721
1205933119894
120593119879
2119894minus119876 0 0
119872119879
10 119872
20
120593119879
31198940 0 minus120576
119894119868
]]]]]]]]]
]
lt 0 (43)
Case 3 If 119894 isin 119880119894
119896and 119880
119894
119906119896= 0 a set of positive definite
matrices119882119894119895isin R119899times119899 (119894 119895 isin 119880
119894
119896) exist such that
119864119879
119891119894119875119894= 119875119879
119894119864119891119894ge 0 (44)
[[[[[[[[[
[
1205951119894
1205952119894
1198711
1205953119894
120595119879
2119894minus119876 0 0
119871119879
10 1198712
0
120595119879
31198940 0 minus120576
119894119868
]]]]]]]]]
]
lt 0 (45)
where
1206011119894= 1205931119894= 1205951119894
= [1198751119894119860119894+ 119860119879
1198941198751119894
119860119879
1198941198752119894minus 119884119879
1119894minus 119862119879
119894119884119879
2119894
1198752119894119860119894minus 1198841119894minus 1198842119894119862119894
119884119879
1119894+ 1198841119894
]
+ 119876 + 119862119879
119891119862119891+ sum
119895isin119880119894
119896
119894119895119864119879
119891(119875119895minus 119875119894) + sum
119895isin119880119894
119896
1
41205902
119894119895119879119894119895
1206012119894= 1205932119894= 1205952119894= [
1198751119894119860 i 0
1198752119894119860119894minus 1198841119894minus 11988421198941198621198940]
1206013119894= 1205933119894= 1205953119894= [
11987511198941198721119894
11987521198941198721119894minus 11988411198941198721119894minus 11988421198941198722119894
]
1198731= [119864119879
119891(119875119896119894
1
minus 119875119894) 119864119879
119891(119875119896119894
2
minus 119875119894) 119864
119879
119891(119875119896119894
119898
minus 119875119894)]
1198732= diag minus119879
119894119896119894
1
minus119879119894119896119894
119898
1198721= [119864119879
119891(119875119896119894
1
minus 119875119894) 119864119879
119891(119875119896119894
2
minus 119875119894) 119864
119879
119891(119875119896119894
119898
minus 119875119894)]
1198722= diag minus119881
119894119896119894
1119897 minus119881
119894119896119894
119898119897
1198711= [119864119879
119891(1198751minus 119875119894) 119864
119879
119891(119875119894minus1
minus 119875119894)
119864119879
119891(119875119894+1
minus 119875119894) 119864
119879
119891(119875119904minus 119875119894)]
1198712= diag minus119882
1198941 minus119882
119894119904
(46)
Abstract and Applied Analysis 7
Then a suitable robust guaranteed cost observer in theform of (11) has parameters as follows
1198701119894= 119875minus1
11198941198841119894 119870
2119894= 119875minus1
21198941198842119894
(47)
and 119869lowast is a robust guaranteed cost for system (13) with GUTRMatrix (4)
Proof Define
1198601
119894=
[[[[
[
119860119879
119891119894119875119894+ 119875119894119860119891119894+ 119876 + sum
119895isin119880119894
119896
119894119895(119875119895minus 119875119894) + sum
119895isin119880119894
119896
1
41205902
119894119895119879119894119895+ 119862119879
119891119862119891
119875119894119860119891119889119894
1198731
119860119879
119891119889119894119875119894
minus119876 0
lowast lowast 1198732
]]]]
]
(48)
1198602
119894=
[[[[
[
119860119879
119891119894119875119894+ 119875119894119860119891119894+ 119876 + sum
119895isin119880119894
119896
119894119895(119875119895minus 119875119894) + sum
119895isin119880119894
119896
1
41205902
119894119895119881119894119895119897+ 119862119879
119891119862119891
119875119894119860119891119889119894
1198721
119860119879
119891119889119894119875119894
minus119876 0
lowast lowast 1198722
]]]]
]
(49)
1198603
119894=
[[[[
[
119860119879
119891119894119875119894+ 119875119894119860119891119894+ 119876 + sum
119895isin119880119894
119896
119894119895(119875119895minus 119875119894) + sum
119895isin119880119894
119896
1
41205902
119894119895119882119894119895+ 119862119879
119891119862119891
119875119894119860119891119889119894
1198711
119860119879
119891119889119894119875119894
minus119876 0
lowast lowast 1198712
]]]]
]
lt 0 (50)
Then (16) is equivalent to
1198601
119894+ [
[
119875119894119872119891119894
0
0
]
]
119865119894[119873119891119894 1198731198911119894 0]
+ [119873119891119894 1198651198911119894 0]119879
119865119879
119894[
[
119875119894119872119891119894
0
0
]
]
119879
lt 0
(51)
Then (19) is equivalent to
1198602
119894+ [
[
119875119894119872119891119894
0
0
]
]
119865119894[119873119891119894 1198731198911119894 0]
+ [119873119891119894 1198651198911119894 0]119879
119865119879
119894[
[
119875119894119872119891119894
0
0
]
]
119879
lt 0
(52)
Then (21) is equivalent to
1198603
119894+ [
[
119875119894119872119891119894
0
0
]
]
119865119894[119873119891119894 1198731198911119894 0]
+ [119873119891119894 1198651198911119894 0]119879
119865119879
119894[
[
119875119894119872119891119894
0
0
]
]
119879
lt 0
(53)
By applying Lemma 24 in [57] (50) (51) and (52) hold forall uncertainties119865i satisfying119865
119879
119894119865119894lt 119868 if and only if there exist
positive scalars 120576119894 119894 = 1 2 119904 such that
1198601
119894+ 120576minus1
119894[
[
119875119894119872119891119894
0
0
]
]
[
[
119875119894119872119891119894
0
0
]
]
119879
+ 120576119894[119873119891119894 1198651198911119894 0]
119879
[119873119891119894 1198651198911119894 0] lt 0
1198602
119894+ 120576minus1
119894[
[
119875119894119872119891119894
0
0
]
]
[
[
119875119894119872119891119894
0
0
]
]
119879
+ 120576119894[119873119891119894 1198651198911119894 0]
119879
[119873119891119894 1198651198911119894 0] lt 0
1198603
119894+ 120576minus1
119894[
[
119875119894119872119891119894
0
0
]
]
[
[
119875119894119872119891119894
0
0
]
]
119879
+ 120576119894[119873119891119894 1198651198911119894 0]
119879
[119873119891119894 1198651198911119894 0] lt 0
(54)
Let 119875119894= diag(119875
1119894 1198752119894) and using (47) we can conclude from
Schur complement results that the above matrix inequalitiesare equivalent to the coupled LMIs (40) (43) and (45)It further follows from Theorem 10 that 119869
lowast is a robustguaranteed cost for system (13) with (4)
Remark 12 The solution of LMIs (39)ndash(45) parameterizesthe set of the proposed robust guaranteed cost observersThis parameterized representation can be used to design theguaranteed cost observer with some additional performanceconstraints By applying the methods in [14] the suboptimal
8 Abstract and Applied Analysis
guaranteed cost observer can be determined by solving acertain optimization problem This is the following theorem
Theorem 13 Consider system (13)with GUTRMatrix (4) andthe cost function (14) and suppose that the initial conditions 119903
0
and 1199091198910
are known if the following optimization problem
min1198761198751119894 1198752119894 120576119894 1198841119894 and 1198842119894
119869lowast
st LMIs (39) ndash (45)(55)
has a solution 119876 1198751119894 1198752119894 120576119894 1198841119894 and 119884
2119894 119894 = 1 2 119904
then the observer (11) is a suboptimal guaranteed costobserver for system (1) where 119869
lowast= 119909119879
1198910119864119879
1198911199030119875(1199030)1199091198910
+
tr(int0minus119889
1199091198910(119905)1199091198910(119905)119909119879
1198910119889119905119876)
Proof It follows from Theorem 11 that the observer (11)constructed in terms of the solution 119876 119875
1119894 1198752119894 120576119894 1198841119894 and
1198842119894 119894 = 1 2 119904 is a robust guaranteed cost observer By
noting that
int
0
minus119889
119909119879
1198910(119905) 119876119909
1198910(119905) 119889119905 = int
0
minus119889
tr (1199091198791198910
(119905) 1198761199091198910
(119905)) 119889119905
= tr(int0
minus119889
119909119879
1198910(119905) 1199091198910
(119905) 119889119905119876)
(56)
it follows that the suboptimal guaranteed cost observerproblem is turned into the minimization problem (55)
Remark 14 Theorem 13 gives the suboptimal guaranteed costobserver conditions of a class of linear Markovian jump-ing time-delay systems with generally incomplete transitionprobability and LMI constraints which can be easily solvedby the LMI toolbox in MATLAB
4 Numerical Example
In this section a numerical example is presented todemonstrate the effectiveness of the method mentioned inTheorem 11 Consider a 2-dimensional system (1) with 3Markovian switching modes In this numerical example thesingular system matrix is set as 119864 = [ 1 0
0 0] and the 3-mode
transition rate matrix is Λ = [minus32
2
15 21 minus36
] where Δ11 Δ31
isin
[minus015 015] Δ23 Δ33
isin [minus012 012] and Δ32
isin [minus01 01]The other system matrices are as follows
For mode 119894 = 1 there are
1198601= [
minus32 065
1 02] 119860
1198891= [
02 05
1 minus068]
1198621= [
[
12 065
minus65 19
minus021 minus18
]
]
1198621198891
= [
[
minus36 minus105
21 096
021 minus086
]
]
11987211
= [minus02
08] 119872
21= [
[
025
0875
minus2
]
]
11987311
= [minus12 31] 11987321
= [minus069 minus42]
(57)
For mode 119894 = 2 there are
1198602= [
minus1 6
2 minus36] 119860
1198892= [
minus31 minus16
3 075]
1198622= [
[
9 minus25
035 minus2
36 minus18
]
]
1198621198892
= [
[
089 minus6
minus12 09
minus24 6
]
]
11987212
= [23
minus4] 119872
22= [
[
075
minus36
25
]
]
11987312
= [minus72 minus6] 11987322
= [1 2]
(58)
For mode 119894 = 3 there are
1198603= [
minus106 29
minus03 36] 119860
1198893= [
minus56 minus12
minus3 45]
1198623= [
[
minus3 minus036
015 minus18
09 minus5
]
]
1198621198893
= [
[
minus165 5
minus12 265
minus098 minus56
]
]
11987213
= [minus82
minus03] 119872
23= [
[
minus052
25
minus36
]
]
11987313
= [105 minus5] 11987323
= [minus72 minus126]
(59)
Then we set the error state matrix 119871 = [ minus45 062 minus6
] and thepositive scalars in Theorem 11 are 120576
1= 02 120576
2= 015 120576
3=
032 According to the definitions of augmented statematricesin (12) we can easily obtain the following parameter matricesin Theorem 11 by MATLAB
11988411
= [minus84521006 00127
00127 84509001]
11988421
= [002 01291 01435
minus23520 minus04080 minus08106]
11988412
= [minus170991 269626
269626 minus246750]
11988422
= [minus200744 minus132941 529388
216893 180120 minus506693]
11988413
= [minus6751329 224456
224456 minus8976976]
11988423
= [minus136021 minus1461726 540500
minus34324 minus195125 minus103068]
1198751=[[[
[
63029 0 0 0
0 48620 0 0
0 0 22914 0
0 0 0 03169
]]]
]
Abstract and Applied Analysis 9
1198752=[[[
[
08914 0 0 0
0 12505 0 0
0 0 73629 0
0 0 0 30056
]]]
]
1198753=[[[
[
30265 0 0 0
0 02156 0 0
0 0 08965 0
0 0 0 10002
]]]
]
119876 =[[[
[
05000 0 0 0
0 05001 0 0
0 0 05001 0
0 0 0 05001
]]]
]
11987911
=[[[
[
34173214 minus8707765 0 0
minus8707765 4167216 0 0
0 0 22263598 minus3207456
0 0 minus3207456 12263101
]]]
]
11987923
=[[[
[
37753231 minus27999330 0 0
minus27999330 28107685 0 0
0 0 106907366 minus107432750
0 0 minus107432750 108555053
]]]
]
11987931
=[[[
[
9518504 minus5399245 0 0
minus5399245 8962029 0 0
0 0 14773012 minus2077540
0 0 minus2077540 14791256
]]]
]
11987932
=[[[
[
21617695 minus12094164 0 0
minus12094164 20371205 0 0
0 0 14786 minus09283
0 0 minus09283 14794
]]]
]
11987933
=[[[
[
14939313 minus8398780 0 0
minus8398780 14073689 0 0
0 0 1478123 minus1336452
0 0 minus1336452 2459347
]]]
]
11988111
=[[[
[
16650 0 0 0
0 16650 0 0
0 0 16650 0
0 0 0 16650
]]]
]
11988231
=[[[
[
15426 0 0 0
0 16650 0 0
0 0 16662 0
0 0 0 16650
]]]
]
11988232
=[[[
[
15428 0 0 0
0 16650 0 0
0 0 16622 0
0 0 0 16650
]]]
]
(60)
Therefore we can design a linear memoryless observer as(11) with the constant matrices
11987011
= 119875minus1
1111988411
= [minus13409860 00020
00026 17381530]
11987021
= 119875minus1
2111988421
= [00087 00563 00626
minus74219 minus12875 minus25579]
11987012
= 119875minus1
1211988412
= [minus191823 302475
215615 minus197321]
11987022
= 119875minus1
2211988422
= [minus27264 minus18056 71899
72163 59928 minus168583]
11987013
= 119875minus1
1311988413
= [minus2231402 74186
1041076 minus41637180]
11987023
= 119875minus1
2311988423
= [minus151724 minus1630481 602900
minus34317 minus195086 minus103047]
(61)
Finally the observer (11) with the above parameter matri-ces for this numerical example is a suboptimal guaranteedcost observer byTheorems 11 and 13
5 Conclusions
In this paper the robust guaranteed cost observer problemfor a class of uncertain descriptor time-delay systems withMarkovian jumping parameters and generally uncertaintransition rates is studied by using LMI method In thisGUTR singular model each transition rate can be com-pletely unknown or only its estimate value is known Theparameterrsquos uncertainty is time varying and is assumed tobe norm-bounded Memoryless guaranteed cost observersare designed in terms of a set of linear coupled matrixinequalities The suboptimal guaranteed cost observer isdesigned by solving a certain optimization problem Ourresults can be applicable to the general Markovian jumpsystems with bounded uncertain or partly uncertain TRmatrix
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This research is supported by the National NaturalScience Foundations of China (10701020 1097102260973048 61272077 60974025 60673101 and 60939003)NCET-08-0755 National 863 Plan Project (2008AA04Z401 2009AA043404) SDPW IMZQWH010016the Natural Science Foundation of Shandong Province (noZR2012FM006) the Scientific and Technological Projectof Shandong Province (no 2007GG3WZ04016) and theNatural Science Foundation of Guangxi Autonomous Region(no 2012GXNSFBA053003)
10 Abstract and Applied Analysis
References
[1] L Dai Singular Control Systems Springer Berlin Germany1989
[2] S Xu P V Dooren R Stefan and J Lam ldquoRobust stability andstabilization for singular systemswith state delay and parameteruncertaintyrdquo IEEE Transactions on Automatic Control vol 47no 7 pp 1122ndash1128 2002
[3] M Fu and G Duan ldquoRobust guaranteed cost observer foruncertain descriptor time-delay systems withMarkovian jump-ing parametersrdquo Acta Automatica Sinica vol 31 pp 479ndash4832005
[4] Y Q Xia E-K Boukas P Shi and J H Zhang ldquoStabilityand stabilization of continuous-time singular hybrid systemsrdquoAutomatica vol 45 no 6 pp 1504ndash1509 2009
[5] S Xu and J Lam Robust Control and Filtering of SingularSystems Springer Berlin Germany 2006
[6] Y-Y Cao and J Lam ldquoRobust 119867infin
control of uncertainMarkovian jump systems with time-delayrdquo IEEE Transactionson Automatic Control vol 45 no 1 pp 77ndash83 2000
[7] Y Li and Y Kao ldquoStability of stochastic reaction-diffusion sys-tems with Markovian switching and impulsive perturbationsrdquoMathematical Problems in Engineering vol 2012 Article ID429568 13 pages 2012
[8] X Mao and C Yuan Stochastic Differential Equations withMarkovian Switching Imperial College Press London UK2006
[9] E-K Boukas Stochastic Switching Systems Analysis and DesignBirkhauser Basel Switzerland 2005
[10] Y Kao C Wang and L Zhang ldquoDelay-dependent exponentialstability of impulsive markovian jumping Cohen-Grossbergneural networks with reaction-diffusion and mixed delaysrdquoNeural Processing Letters vol 38 no 3 pp 321ndash346 2013
[11] Y-G Kao J-F Guo C-H Wang and X-Q Sun ldquoDelay-dependent robust exponential stability of Markovian jump-ing reaction-diffusion Cohen-Grossberg neural networks withmixed delaysrdquo Journal of the Franklin Institute vol 349 no 6pp 1972ndash1988 2012
[12] Y Kao C Wang F Zha and H Cao ldquoStability in mean ofpartial variables for stochastic reactionmdashdiffusion systems withMarkovian switchingrdquo Journal of the Franklin Institute vol 351no 1 pp 500ndash512 2014
[13] N Kumaresan and P Balasubramaniam ldquoOptimal control forstochastic nonlinear singular system using neural networksrdquoComputers amp Mathematics with Applications vol 56 no 9 pp2145ndash2154 2008
[14] L Yu and J Chu ldquoAn LMI approach to guaranteed cost controlof linear uncertain time-delay systemsrdquoAutomatica vol 35 no6 pp 1155ndash1159 1999
[15] S Yin H Luo and S Ding ldquoReal-time implementation of fault-tolerant control systems with performance optimizationrdquo IEEETransactions on Industrial Electronics vol 64 no 5 pp 2402ndash2411 2014
[16] S Yin G Wang and H Karimi ldquoData-driven design of robustfault detection system for wind turbinesrdquoMechatronics 2013
[17] S Yin S X Ding A H A Sari and H Hao ldquoData-drivenmonitoring for stochastic systems and its application on batchprocessrdquo International Journal of Systems Science vol 44 no 7pp 1366ndash1376 2013
[18] S Yin S Ding A Haghani H Hao and P Zhang ldquoA com-parison study of basic datadriven fault diagnosis and process
monitoring methods on the benchmark Tennessee Eastmanprocessrdquo Journal of Process Control vol 22 no 9 pp 1567ndash15812012
[19] Y G Kao and C H Wang ldquoGlobal stability analysis forstochastic coupled reaction-diffusion systems on networksrdquoNonlinear Analysis Real World Applications vol 14 no 3 pp1457ndash1465 2013
[20] E K Boukas ldquoStabilization of stochastic singular nonlinearhybrid systemsrdquo Nonlinear Analysis Theory Methods amp Appli-cations vol 64 no 2 pp 217ndash228 2006
[21] E-K Boukas Control of Singular Systems with Random AbruptChanges Springer Berlin Germany 2008
[22] Y Ding H Zhu S Zhong and Y Zeng ldquoExponential mean-square stability of time-delay singular systems with Markovianswitching and nonlinear perturbationsrdquo Applied Mathematicsand Computation vol 219 no 4 pp 2350ndash2359 2012
[23] S Long S Zhong and Z Liu ldquoStochastic admissibility fora class of singular Markovian jump systems with mode-dependent time delaysrdquoAppliedMathematics and Computationvol 219 no 8 pp 4106ndash4117 2012
[24] Z Wu P Shi H Su and J Chu ldquoAsynchronous 1198682minus 119868infin
filtering for discrete-time stochastic Markov jump systems withrandomly occurred sensor nonlinearitiesrdquo Automatica vol 50no 1 pp 180ndash186 2014
[25] Z Wu P Shi H Su and J Chu ldquoStochastic synchronizationof Markovian jump neural networks with time-varying delayusing sampled-datardquo IEEE Transactions on Cybernetics vol 43no 6 pp 1796ndash1806 2013
[26] S Ma and E-K Boukas ldquoGuaranteed cost control of uncertaindiscrete-time singular Markov jump systems with indefinitequadratic costrdquo International Journal of Robust and NonlinearControl vol 21 no 9 pp 1031ndash1045 2011
[27] G Wang and Q Zhang ldquoRobust control of uncertain singularstochastic systems with Markovian switching via proportional-derivative state feedbackrdquo IET Control Theory amp Applicationsvol 6 no 8 pp 1089ndash1096 2012
[28] J Wang H Wang A Xue and R Lu ldquoDelay-dependent 119867infin
control for singular Markovian jump systems with time delayrdquoNonlinear Analysis Hybrid Systems vol 8 pp 1ndash12 2013
[29] L Wu X Su and P Shi ldquoSliding mode control with bounded1198712gain performance of Markovian jump singular time-delay
systemsrdquo Automatica vol 48 no 8 pp 1929ndash1933 2012[30] Z-GWu J H ParkH Su and J Chu ldquoStochastic stability anal-
ysis for discrete-time singular Markov jump systems with time-varying delay and piecewise-constant transition probabilitiesrdquoJournal of the Franklin Institute vol 349 no 9 pp 2889ndash29022012
[31] J Xiong J Lam H Gao and D W C Ho ldquoOn robust stabi-lization of Markovian jump systems with uncertain switchingprobabilitiesrdquo Automatica vol 41 no 5 pp 897ndash903 2005
[32] J Xiong and J Lam ldquoRobust 1198672control of Markovian jump
systems with uncertain switching probabilitiesrdquo InternationalJournal of Systems Science vol 40 no 3 pp 255ndash265 2009
[33] M Karan P Shi and C Y Kaya ldquoTransition probabilitybounds for the stochastic stability robustness of continuous-and discrete-timeMarkovian jump linear systemsrdquoAutomaticavol 42 no 12 pp 2159ndash2168 2006
[34] L Zhang and E-K Boukas ldquoStability and stabilization ofMarkovian jump linear systems with partly unknown transitionprobabilitiesrdquo Automatica vol 45 no 2 pp 463ndash468 2009
Abstract and Applied Analysis 11
[35] B Du J Lam Y Zou and Z Shu ldquoStability and stabilization forMarkovian jump time-delay systems with partially unknowntransition ratesrdquo IEEE Transactions on Circuits and Systems IRegular Papers vol 60 no 2 pp 341ndash351 2013
[36] L Zhang and J Lam ldquoNecessary and sufficient conditionsfor analysis and synthesis of Markov jump linear systemswith incomplete transition descriptionsrdquo IEEE Transactions onAutomatic Control vol 55 no 7 pp 1695ndash1701 2010
[37] Y Guo and F Zhu ldquoNew results on stability and stabilizationof Markovian jump systems with partly known transitionprobabilitiesrdquoMathematical Problems in Engineering vol 2012Article ID 869842 11 pages 2012
[38] J Lin S Fei and J Shen ldquoDelay-dependent 119867infin
filtering fordiscrete-time singular Markovian jump systems with time-varying delay and partially unknown transition probabilitiesrdquoSignal Processing vol 91 no 2 pp 277ndash289 2011
[39] J Tian Y Li J Zhao and S Zhong ldquoDelay-dependent stochasticstability criteria for Markovian jumping neural networks withmode-dependent time-varying delays and partially knowntransition ratesrdquo Applied Mathematics and Computation vol218 no 9 pp 5769ndash5781 2012
[40] Y Wei J Qiu H R Karimi and M Wang ldquoA new design 119867infin
filtering for contrinuous-time Markovian jump systems withtime-varying delay and partially accessible mode informationrdquoSignal Processing vol 93 pp 2392ndash2407 2013
[41] L Zhang E-K Boukas and J Lam ldquoAnalysis and synthesisof Markov jump linear systems with time-varying delays andpartially known transition probabilitiesrdquo IEEE Transactions onAutomatic Control vol 53 no 10 pp 2458ndash2464 2008
[42] L Zhang and E-K Boukas ldquoMode-dependent 119867infin
filteringfor discrete-time Markovian jump linear systems with partlyunknown transition probabilitiesrdquo Automatica vol 45 no 6pp 1462ndash1467 2009
[43] L Zhang and E-K Boukas ldquoMode-dependent 119867infin
controlfor discrete-time Markovian jump linear systems with partlyunknown transition probabilitiesrdquo International Journal ofRobust and Nonlinear Control vol 19 no 8 pp 868ndash883 2009
[44] Y Zhang Y He M Wu and J Zhang ldquoStabilization forMarkovian jump systems with partial information on transi-tion probability based on free-connection weighting matricesrdquoAutomatica vol 47 no 1 pp 79ndash84 2011
[45] G Zong D Yang L Hou and Q Wang ldquoRobust finite-time119867infincontrol for Markovian jump systems with partially known
transition probabilitiesrdquo Journal of the Franklin Institute vol350 no 6 pp 1562ndash1578 2013
[46] Y Ding H Zhu S Zhong and Y Zhang ldquo1198712minus 119871infin
filteringfor Markovian jump systems with time-varying delays andpartly unknown transition probabilitiesrdquo Communications inNonlinear Science and Numerical Simulation vol 17 no 7 pp3070ndash3081 2012
[47] Q Ma S Xu and Y Zou ldquoStability and synchronizationfor Markovian jump neural networks with partly unknowntransition probabilitiesrdquo Neurocomputing vol 74 pp 3403ndash3411 2011
[48] E Tian D Yue and G Wei ldquoRobust control for Markovianjump systems with partially known transition probabilities andnonlinearitiesrdquo Journal of the Franklin Institute vol 350 no 8pp 2069ndash2083 2013
[49] Y Guo and Z Wang ldquoStability of Markovian jump systemswith generally uncertain transition ratesrdquo Journal of the FranklinInstitute vol 350 no 9 pp 2826ndash2836 2013
[50] L Zhang ldquo119867infinestimation for discrete-time piecewise homoge-
neous Markov jump linear systemsrdquo Automatica vol 45 no 11pp 2570ndash2576 2009
[51] Z Wu H Su and J Chu ldquostate estimation for discreteMarkovian jumping neural networks with time delayrdquo Neuro-computing vol 73 pp 2247ndash2254 2010
[52] I R Petersen and A V Savkin Robust Kalman Filteringfor Signals and Systems with Large Uncertainties BirkhauserBoston Mass USA 1999
[53] B D O Anderson and J B Moore Optimal Filtering Prentice-Hall Upper Saddle River NJ USA 1979
[54] I R Petersen and D C McFarlane ldquoOptimal guaranteedcost control and filtering for uncertain linear systemsrdquo IEEETransactions on Automatic Control vol 39 no 9 pp 1971ndash19771994
[55] S Xu T Chen and J Lam ldquoRobust 119867infin filtering for uncertainMarkovian jump systems with mode-dependent time delaysrdquoIEEE Transactions on Automatic Control vol 48 no 5 pp 900ndash907 2003
[56] J H Kim D Oh and H Park ldquoGuaranteed cost and 119867infin
filtering for time-delay systemsrdquo in Proceeding of the AmericanControl Conference pp 4014ndash4018 IEEE Arlington Va USA2001
[57] L Xie ldquoOutput feedback119867infincontrol of systems with parameter
uncertaintyrdquo International Journal of Control vol 63 no 4 pp741ndash750 1996
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
6 Abstract and Applied Analysis
Case 3 Substituting (29) into (24) we get
L119881 le Λ119879(119905) Γ (119894) Λ (119905) (34)
where Λ119879(119905) = [119909119879
119891(119905) 119909119879
119891(119905 minus 119889)] and
Γ119894=[[
[
Δ119894+ 119862119879
119891119862119891
119875119894(119860119891119889119894
+ Δ119860119891119889119894
) 1
(119860119891119889119894
+ Δ119860119891119889119894
)119879
119875119894
minus119876 0
lowast lowast 2
]]
]
(35)
Similar to [5] usingDynkinrsquos formula we drive for each 119894 isin 119873
lim119879rarrinfin
Eint
119879
0
119909119879
119891(119905) 119909119891(119905) 119889119905 | 120593
119891 1205740= 119894 le 119909
119879
11989101198721199091198910 (36)
By Definition 5 it is easy to see that the augmented system(13) is stochastically stable Furthermore from (16) (19) and(21) we have
L119881 le minus119909119879
119891(119905) 119862119879
119891119862119891119909119891(119905) lt 0 (37)
On the other hand we have
J = Eint
infin
0
119909119879
119891(119905) 119862119879
119891119862119891119909119891(119905) 119889119905 lt minusint
infin
0
L119881119889119905
= minus E lim119905rarrinfin
119881 (119909 (119905) 120574 (119905)) + 119881 (1199090 1205740)
(38)
As the augmented system (13) is stochastically stable itfollows from (38) that 119869 lt 119881(119909
1198910 1199030) From Definition 7 it
is concluded that a robust guaranteed cost for the augmentedsystem (13) can be given by 119869
lowast= 119909119879
1198910(119905)119864119879
1198911199030119875(1199030)1199091198910
+
int0
minus119889119909119879
119891(119905)119876119909
119891(119905)119889119905
In the following based on the above sufficient conditionthe design of robust guaranteed cost observers can be turnedinto the solvability of a system of LMIs
Theorem 11 Consider system (13)with GUTRMatrix (4) andthe cost function (14) If there exist matrices 119884
1119894and 119884
2119894 119894 =
1 2 119904 positive scalars 120576119894 119894 = 1 2 119904 symmetric positive
definite matrix Q and the full rank matrices 1198752119894 and matrices
119875119894= diag(119875
1119894 1198752119894) 119894 = 1 2 119904 satisfying the following
LMIs respectively
Case 1 If 119894 notin 119880119894
119896and 119880
119894
119896= 119896119894
1 119896
119894
119898 a set of positive
definite matrices 119879119894119895isin R119899times119899 (119894 notin 119880
119894
119896 119895 isin 119880
119894
119896) exist such that
119864119879
119891119875119894= 119875119879
119894119864119891ge 0 (39)
[[[[[[[[[
[
1206011119894
1206012119894
1198731
1206013119894
120601119879
2119894minus119876 0 0
119873119879
10 119873
20
120601119879
31198940 0 minus120576
119894119868
]]]]]]]]]
]
lt 0 (40)
119875119894minus 119875119895ge 0 forall119895 isin 119880
119894
119906119896 119895 = 119894 (41)
Case 2 If 119894 isin 119880119894
119896(119880119894
119896= 119896119894
1 119896
119894
119898) and 119880
119894
119906119896= 0 a set of
positive definite matrices119881119894119895119897isin R119899times119899 (119894 119895 isin 119880
119894
119896 119897 isin 119880
119894
119906119896) exist
such that
119864119879
119891119875119894= 119875119879
119894119864119891ge 0 (42)
[[[[[[[[[
[
1205931119894
1205932119894
1198721
1205933119894
120593119879
2119894minus119876 0 0
119872119879
10 119872
20
120593119879
31198940 0 minus120576
119894119868
]]]]]]]]]
]
lt 0 (43)
Case 3 If 119894 isin 119880119894
119896and 119880
119894
119906119896= 0 a set of positive definite
matrices119882119894119895isin R119899times119899 (119894 119895 isin 119880
119894
119896) exist such that
119864119879
119891119894119875119894= 119875119879
119894119864119891119894ge 0 (44)
[[[[[[[[[
[
1205951119894
1205952119894
1198711
1205953119894
120595119879
2119894minus119876 0 0
119871119879
10 1198712
0
120595119879
31198940 0 minus120576
119894119868
]]]]]]]]]
]
lt 0 (45)
where
1206011119894= 1205931119894= 1205951119894
= [1198751119894119860119894+ 119860119879
1198941198751119894
119860119879
1198941198752119894minus 119884119879
1119894minus 119862119879
119894119884119879
2119894
1198752119894119860119894minus 1198841119894minus 1198842119894119862119894
119884119879
1119894+ 1198841119894
]
+ 119876 + 119862119879
119891119862119891+ sum
119895isin119880119894
119896
119894119895119864119879
119891(119875119895minus 119875119894) + sum
119895isin119880119894
119896
1
41205902
119894119895119879119894119895
1206012119894= 1205932119894= 1205952119894= [
1198751119894119860 i 0
1198752119894119860119894minus 1198841119894minus 11988421198941198621198940]
1206013119894= 1205933119894= 1205953119894= [
11987511198941198721119894
11987521198941198721119894minus 11988411198941198721119894minus 11988421198941198722119894
]
1198731= [119864119879
119891(119875119896119894
1
minus 119875119894) 119864119879
119891(119875119896119894
2
minus 119875119894) 119864
119879
119891(119875119896119894
119898
minus 119875119894)]
1198732= diag minus119879
119894119896119894
1
minus119879119894119896119894
119898
1198721= [119864119879
119891(119875119896119894
1
minus 119875119894) 119864119879
119891(119875119896119894
2
minus 119875119894) 119864
119879
119891(119875119896119894
119898
minus 119875119894)]
1198722= diag minus119881
119894119896119894
1119897 minus119881
119894119896119894
119898119897
1198711= [119864119879
119891(1198751minus 119875119894) 119864
119879
119891(119875119894minus1
minus 119875119894)
119864119879
119891(119875119894+1
minus 119875119894) 119864
119879
119891(119875119904minus 119875119894)]
1198712= diag minus119882
1198941 minus119882
119894119904
(46)
Abstract and Applied Analysis 7
Then a suitable robust guaranteed cost observer in theform of (11) has parameters as follows
1198701119894= 119875minus1
11198941198841119894 119870
2119894= 119875minus1
21198941198842119894
(47)
and 119869lowast is a robust guaranteed cost for system (13) with GUTRMatrix (4)
Proof Define
1198601
119894=
[[[[
[
119860119879
119891119894119875119894+ 119875119894119860119891119894+ 119876 + sum
119895isin119880119894
119896
119894119895(119875119895minus 119875119894) + sum
119895isin119880119894
119896
1
41205902
119894119895119879119894119895+ 119862119879
119891119862119891
119875119894119860119891119889119894
1198731
119860119879
119891119889119894119875119894
minus119876 0
lowast lowast 1198732
]]]]
]
(48)
1198602
119894=
[[[[
[
119860119879
119891119894119875119894+ 119875119894119860119891119894+ 119876 + sum
119895isin119880119894
119896
119894119895(119875119895minus 119875119894) + sum
119895isin119880119894
119896
1
41205902
119894119895119881119894119895119897+ 119862119879
119891119862119891
119875119894119860119891119889119894
1198721
119860119879
119891119889119894119875119894
minus119876 0
lowast lowast 1198722
]]]]
]
(49)
1198603
119894=
[[[[
[
119860119879
119891119894119875119894+ 119875119894119860119891119894+ 119876 + sum
119895isin119880119894
119896
119894119895(119875119895minus 119875119894) + sum
119895isin119880119894
119896
1
41205902
119894119895119882119894119895+ 119862119879
119891119862119891
119875119894119860119891119889119894
1198711
119860119879
119891119889119894119875119894
minus119876 0
lowast lowast 1198712
]]]]
]
lt 0 (50)
Then (16) is equivalent to
1198601
119894+ [
[
119875119894119872119891119894
0
0
]
]
119865119894[119873119891119894 1198731198911119894 0]
+ [119873119891119894 1198651198911119894 0]119879
119865119879
119894[
[
119875119894119872119891119894
0
0
]
]
119879
lt 0
(51)
Then (19) is equivalent to
1198602
119894+ [
[
119875119894119872119891119894
0
0
]
]
119865119894[119873119891119894 1198731198911119894 0]
+ [119873119891119894 1198651198911119894 0]119879
119865119879
119894[
[
119875119894119872119891119894
0
0
]
]
119879
lt 0
(52)
Then (21) is equivalent to
1198603
119894+ [
[
119875119894119872119891119894
0
0
]
]
119865119894[119873119891119894 1198731198911119894 0]
+ [119873119891119894 1198651198911119894 0]119879
119865119879
119894[
[
119875119894119872119891119894
0
0
]
]
119879
lt 0
(53)
By applying Lemma 24 in [57] (50) (51) and (52) hold forall uncertainties119865i satisfying119865
119879
119894119865119894lt 119868 if and only if there exist
positive scalars 120576119894 119894 = 1 2 119904 such that
1198601
119894+ 120576minus1
119894[
[
119875119894119872119891119894
0
0
]
]
[
[
119875119894119872119891119894
0
0
]
]
119879
+ 120576119894[119873119891119894 1198651198911119894 0]
119879
[119873119891119894 1198651198911119894 0] lt 0
1198602
119894+ 120576minus1
119894[
[
119875119894119872119891119894
0
0
]
]
[
[
119875119894119872119891119894
0
0
]
]
119879
+ 120576119894[119873119891119894 1198651198911119894 0]
119879
[119873119891119894 1198651198911119894 0] lt 0
1198603
119894+ 120576minus1
119894[
[
119875119894119872119891119894
0
0
]
]
[
[
119875119894119872119891119894
0
0
]
]
119879
+ 120576119894[119873119891119894 1198651198911119894 0]
119879
[119873119891119894 1198651198911119894 0] lt 0
(54)
Let 119875119894= diag(119875
1119894 1198752119894) and using (47) we can conclude from
Schur complement results that the above matrix inequalitiesare equivalent to the coupled LMIs (40) (43) and (45)It further follows from Theorem 10 that 119869
lowast is a robustguaranteed cost for system (13) with (4)
Remark 12 The solution of LMIs (39)ndash(45) parameterizesthe set of the proposed robust guaranteed cost observersThis parameterized representation can be used to design theguaranteed cost observer with some additional performanceconstraints By applying the methods in [14] the suboptimal
8 Abstract and Applied Analysis
guaranteed cost observer can be determined by solving acertain optimization problem This is the following theorem
Theorem 13 Consider system (13)with GUTRMatrix (4) andthe cost function (14) and suppose that the initial conditions 119903
0
and 1199091198910
are known if the following optimization problem
min1198761198751119894 1198752119894 120576119894 1198841119894 and 1198842119894
119869lowast
st LMIs (39) ndash (45)(55)
has a solution 119876 1198751119894 1198752119894 120576119894 1198841119894 and 119884
2119894 119894 = 1 2 119904
then the observer (11) is a suboptimal guaranteed costobserver for system (1) where 119869
lowast= 119909119879
1198910119864119879
1198911199030119875(1199030)1199091198910
+
tr(int0minus119889
1199091198910(119905)1199091198910(119905)119909119879
1198910119889119905119876)
Proof It follows from Theorem 11 that the observer (11)constructed in terms of the solution 119876 119875
1119894 1198752119894 120576119894 1198841119894 and
1198842119894 119894 = 1 2 119904 is a robust guaranteed cost observer By
noting that
int
0
minus119889
119909119879
1198910(119905) 119876119909
1198910(119905) 119889119905 = int
0
minus119889
tr (1199091198791198910
(119905) 1198761199091198910
(119905)) 119889119905
= tr(int0
minus119889
119909119879
1198910(119905) 1199091198910
(119905) 119889119905119876)
(56)
it follows that the suboptimal guaranteed cost observerproblem is turned into the minimization problem (55)
Remark 14 Theorem 13 gives the suboptimal guaranteed costobserver conditions of a class of linear Markovian jump-ing time-delay systems with generally incomplete transitionprobability and LMI constraints which can be easily solvedby the LMI toolbox in MATLAB
4 Numerical Example
In this section a numerical example is presented todemonstrate the effectiveness of the method mentioned inTheorem 11 Consider a 2-dimensional system (1) with 3Markovian switching modes In this numerical example thesingular system matrix is set as 119864 = [ 1 0
0 0] and the 3-mode
transition rate matrix is Λ = [minus32
2
15 21 minus36
] where Δ11 Δ31
isin
[minus015 015] Δ23 Δ33
isin [minus012 012] and Δ32
isin [minus01 01]The other system matrices are as follows
For mode 119894 = 1 there are
1198601= [
minus32 065
1 02] 119860
1198891= [
02 05
1 minus068]
1198621= [
[
12 065
minus65 19
minus021 minus18
]
]
1198621198891
= [
[
minus36 minus105
21 096
021 minus086
]
]
11987211
= [minus02
08] 119872
21= [
[
025
0875
minus2
]
]
11987311
= [minus12 31] 11987321
= [minus069 minus42]
(57)
For mode 119894 = 2 there are
1198602= [
minus1 6
2 minus36] 119860
1198892= [
minus31 minus16
3 075]
1198622= [
[
9 minus25
035 minus2
36 minus18
]
]
1198621198892
= [
[
089 minus6
minus12 09
minus24 6
]
]
11987212
= [23
minus4] 119872
22= [
[
075
minus36
25
]
]
11987312
= [minus72 minus6] 11987322
= [1 2]
(58)
For mode 119894 = 3 there are
1198603= [
minus106 29
minus03 36] 119860
1198893= [
minus56 minus12
minus3 45]
1198623= [
[
minus3 minus036
015 minus18
09 minus5
]
]
1198621198893
= [
[
minus165 5
minus12 265
minus098 minus56
]
]
11987213
= [minus82
minus03] 119872
23= [
[
minus052
25
minus36
]
]
11987313
= [105 minus5] 11987323
= [minus72 minus126]
(59)
Then we set the error state matrix 119871 = [ minus45 062 minus6
] and thepositive scalars in Theorem 11 are 120576
1= 02 120576
2= 015 120576
3=
032 According to the definitions of augmented statematricesin (12) we can easily obtain the following parameter matricesin Theorem 11 by MATLAB
11988411
= [minus84521006 00127
00127 84509001]
11988421
= [002 01291 01435
minus23520 minus04080 minus08106]
11988412
= [minus170991 269626
269626 minus246750]
11988422
= [minus200744 minus132941 529388
216893 180120 minus506693]
11988413
= [minus6751329 224456
224456 minus8976976]
11988423
= [minus136021 minus1461726 540500
minus34324 minus195125 minus103068]
1198751=[[[
[
63029 0 0 0
0 48620 0 0
0 0 22914 0
0 0 0 03169
]]]
]
Abstract and Applied Analysis 9
1198752=[[[
[
08914 0 0 0
0 12505 0 0
0 0 73629 0
0 0 0 30056
]]]
]
1198753=[[[
[
30265 0 0 0
0 02156 0 0
0 0 08965 0
0 0 0 10002
]]]
]
119876 =[[[
[
05000 0 0 0
0 05001 0 0
0 0 05001 0
0 0 0 05001
]]]
]
11987911
=[[[
[
34173214 minus8707765 0 0
minus8707765 4167216 0 0
0 0 22263598 minus3207456
0 0 minus3207456 12263101
]]]
]
11987923
=[[[
[
37753231 minus27999330 0 0
minus27999330 28107685 0 0
0 0 106907366 minus107432750
0 0 minus107432750 108555053
]]]
]
11987931
=[[[
[
9518504 minus5399245 0 0
minus5399245 8962029 0 0
0 0 14773012 minus2077540
0 0 minus2077540 14791256
]]]
]
11987932
=[[[
[
21617695 minus12094164 0 0
minus12094164 20371205 0 0
0 0 14786 minus09283
0 0 minus09283 14794
]]]
]
11987933
=[[[
[
14939313 minus8398780 0 0
minus8398780 14073689 0 0
0 0 1478123 minus1336452
0 0 minus1336452 2459347
]]]
]
11988111
=[[[
[
16650 0 0 0
0 16650 0 0
0 0 16650 0
0 0 0 16650
]]]
]
11988231
=[[[
[
15426 0 0 0
0 16650 0 0
0 0 16662 0
0 0 0 16650
]]]
]
11988232
=[[[
[
15428 0 0 0
0 16650 0 0
0 0 16622 0
0 0 0 16650
]]]
]
(60)
Therefore we can design a linear memoryless observer as(11) with the constant matrices
11987011
= 119875minus1
1111988411
= [minus13409860 00020
00026 17381530]
11987021
= 119875minus1
2111988421
= [00087 00563 00626
minus74219 minus12875 minus25579]
11987012
= 119875minus1
1211988412
= [minus191823 302475
215615 minus197321]
11987022
= 119875minus1
2211988422
= [minus27264 minus18056 71899
72163 59928 minus168583]
11987013
= 119875minus1
1311988413
= [minus2231402 74186
1041076 minus41637180]
11987023
= 119875minus1
2311988423
= [minus151724 minus1630481 602900
minus34317 minus195086 minus103047]
(61)
Finally the observer (11) with the above parameter matri-ces for this numerical example is a suboptimal guaranteedcost observer byTheorems 11 and 13
5 Conclusions
In this paper the robust guaranteed cost observer problemfor a class of uncertain descriptor time-delay systems withMarkovian jumping parameters and generally uncertaintransition rates is studied by using LMI method In thisGUTR singular model each transition rate can be com-pletely unknown or only its estimate value is known Theparameterrsquos uncertainty is time varying and is assumed tobe norm-bounded Memoryless guaranteed cost observersare designed in terms of a set of linear coupled matrixinequalities The suboptimal guaranteed cost observer isdesigned by solving a certain optimization problem Ourresults can be applicable to the general Markovian jumpsystems with bounded uncertain or partly uncertain TRmatrix
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This research is supported by the National NaturalScience Foundations of China (10701020 1097102260973048 61272077 60974025 60673101 and 60939003)NCET-08-0755 National 863 Plan Project (2008AA04Z401 2009AA043404) SDPW IMZQWH010016the Natural Science Foundation of Shandong Province (noZR2012FM006) the Scientific and Technological Projectof Shandong Province (no 2007GG3WZ04016) and theNatural Science Foundation of Guangxi Autonomous Region(no 2012GXNSFBA053003)
10 Abstract and Applied Analysis
References
[1] L Dai Singular Control Systems Springer Berlin Germany1989
[2] S Xu P V Dooren R Stefan and J Lam ldquoRobust stability andstabilization for singular systemswith state delay and parameteruncertaintyrdquo IEEE Transactions on Automatic Control vol 47no 7 pp 1122ndash1128 2002
[3] M Fu and G Duan ldquoRobust guaranteed cost observer foruncertain descriptor time-delay systems withMarkovian jump-ing parametersrdquo Acta Automatica Sinica vol 31 pp 479ndash4832005
[4] Y Q Xia E-K Boukas P Shi and J H Zhang ldquoStabilityand stabilization of continuous-time singular hybrid systemsrdquoAutomatica vol 45 no 6 pp 1504ndash1509 2009
[5] S Xu and J Lam Robust Control and Filtering of SingularSystems Springer Berlin Germany 2006
[6] Y-Y Cao and J Lam ldquoRobust 119867infin
control of uncertainMarkovian jump systems with time-delayrdquo IEEE Transactionson Automatic Control vol 45 no 1 pp 77ndash83 2000
[7] Y Li and Y Kao ldquoStability of stochastic reaction-diffusion sys-tems with Markovian switching and impulsive perturbationsrdquoMathematical Problems in Engineering vol 2012 Article ID429568 13 pages 2012
[8] X Mao and C Yuan Stochastic Differential Equations withMarkovian Switching Imperial College Press London UK2006
[9] E-K Boukas Stochastic Switching Systems Analysis and DesignBirkhauser Basel Switzerland 2005
[10] Y Kao C Wang and L Zhang ldquoDelay-dependent exponentialstability of impulsive markovian jumping Cohen-Grossbergneural networks with reaction-diffusion and mixed delaysrdquoNeural Processing Letters vol 38 no 3 pp 321ndash346 2013
[11] Y-G Kao J-F Guo C-H Wang and X-Q Sun ldquoDelay-dependent robust exponential stability of Markovian jump-ing reaction-diffusion Cohen-Grossberg neural networks withmixed delaysrdquo Journal of the Franklin Institute vol 349 no 6pp 1972ndash1988 2012
[12] Y Kao C Wang F Zha and H Cao ldquoStability in mean ofpartial variables for stochastic reactionmdashdiffusion systems withMarkovian switchingrdquo Journal of the Franklin Institute vol 351no 1 pp 500ndash512 2014
[13] N Kumaresan and P Balasubramaniam ldquoOptimal control forstochastic nonlinear singular system using neural networksrdquoComputers amp Mathematics with Applications vol 56 no 9 pp2145ndash2154 2008
[14] L Yu and J Chu ldquoAn LMI approach to guaranteed cost controlof linear uncertain time-delay systemsrdquoAutomatica vol 35 no6 pp 1155ndash1159 1999
[15] S Yin H Luo and S Ding ldquoReal-time implementation of fault-tolerant control systems with performance optimizationrdquo IEEETransactions on Industrial Electronics vol 64 no 5 pp 2402ndash2411 2014
[16] S Yin G Wang and H Karimi ldquoData-driven design of robustfault detection system for wind turbinesrdquoMechatronics 2013
[17] S Yin S X Ding A H A Sari and H Hao ldquoData-drivenmonitoring for stochastic systems and its application on batchprocessrdquo International Journal of Systems Science vol 44 no 7pp 1366ndash1376 2013
[18] S Yin S Ding A Haghani H Hao and P Zhang ldquoA com-parison study of basic datadriven fault diagnosis and process
monitoring methods on the benchmark Tennessee Eastmanprocessrdquo Journal of Process Control vol 22 no 9 pp 1567ndash15812012
[19] Y G Kao and C H Wang ldquoGlobal stability analysis forstochastic coupled reaction-diffusion systems on networksrdquoNonlinear Analysis Real World Applications vol 14 no 3 pp1457ndash1465 2013
[20] E K Boukas ldquoStabilization of stochastic singular nonlinearhybrid systemsrdquo Nonlinear Analysis Theory Methods amp Appli-cations vol 64 no 2 pp 217ndash228 2006
[21] E-K Boukas Control of Singular Systems with Random AbruptChanges Springer Berlin Germany 2008
[22] Y Ding H Zhu S Zhong and Y Zeng ldquoExponential mean-square stability of time-delay singular systems with Markovianswitching and nonlinear perturbationsrdquo Applied Mathematicsand Computation vol 219 no 4 pp 2350ndash2359 2012
[23] S Long S Zhong and Z Liu ldquoStochastic admissibility fora class of singular Markovian jump systems with mode-dependent time delaysrdquoAppliedMathematics and Computationvol 219 no 8 pp 4106ndash4117 2012
[24] Z Wu P Shi H Su and J Chu ldquoAsynchronous 1198682minus 119868infin
filtering for discrete-time stochastic Markov jump systems withrandomly occurred sensor nonlinearitiesrdquo Automatica vol 50no 1 pp 180ndash186 2014
[25] Z Wu P Shi H Su and J Chu ldquoStochastic synchronizationof Markovian jump neural networks with time-varying delayusing sampled-datardquo IEEE Transactions on Cybernetics vol 43no 6 pp 1796ndash1806 2013
[26] S Ma and E-K Boukas ldquoGuaranteed cost control of uncertaindiscrete-time singular Markov jump systems with indefinitequadratic costrdquo International Journal of Robust and NonlinearControl vol 21 no 9 pp 1031ndash1045 2011
[27] G Wang and Q Zhang ldquoRobust control of uncertain singularstochastic systems with Markovian switching via proportional-derivative state feedbackrdquo IET Control Theory amp Applicationsvol 6 no 8 pp 1089ndash1096 2012
[28] J Wang H Wang A Xue and R Lu ldquoDelay-dependent 119867infin
control for singular Markovian jump systems with time delayrdquoNonlinear Analysis Hybrid Systems vol 8 pp 1ndash12 2013
[29] L Wu X Su and P Shi ldquoSliding mode control with bounded1198712gain performance of Markovian jump singular time-delay
systemsrdquo Automatica vol 48 no 8 pp 1929ndash1933 2012[30] Z-GWu J H ParkH Su and J Chu ldquoStochastic stability anal-
ysis for discrete-time singular Markov jump systems with time-varying delay and piecewise-constant transition probabilitiesrdquoJournal of the Franklin Institute vol 349 no 9 pp 2889ndash29022012
[31] J Xiong J Lam H Gao and D W C Ho ldquoOn robust stabi-lization of Markovian jump systems with uncertain switchingprobabilitiesrdquo Automatica vol 41 no 5 pp 897ndash903 2005
[32] J Xiong and J Lam ldquoRobust 1198672control of Markovian jump
systems with uncertain switching probabilitiesrdquo InternationalJournal of Systems Science vol 40 no 3 pp 255ndash265 2009
[33] M Karan P Shi and C Y Kaya ldquoTransition probabilitybounds for the stochastic stability robustness of continuous-and discrete-timeMarkovian jump linear systemsrdquoAutomaticavol 42 no 12 pp 2159ndash2168 2006
[34] L Zhang and E-K Boukas ldquoStability and stabilization ofMarkovian jump linear systems with partly unknown transitionprobabilitiesrdquo Automatica vol 45 no 2 pp 463ndash468 2009
Abstract and Applied Analysis 11
[35] B Du J Lam Y Zou and Z Shu ldquoStability and stabilization forMarkovian jump time-delay systems with partially unknowntransition ratesrdquo IEEE Transactions on Circuits and Systems IRegular Papers vol 60 no 2 pp 341ndash351 2013
[36] L Zhang and J Lam ldquoNecessary and sufficient conditionsfor analysis and synthesis of Markov jump linear systemswith incomplete transition descriptionsrdquo IEEE Transactions onAutomatic Control vol 55 no 7 pp 1695ndash1701 2010
[37] Y Guo and F Zhu ldquoNew results on stability and stabilizationof Markovian jump systems with partly known transitionprobabilitiesrdquoMathematical Problems in Engineering vol 2012Article ID 869842 11 pages 2012
[38] J Lin S Fei and J Shen ldquoDelay-dependent 119867infin
filtering fordiscrete-time singular Markovian jump systems with time-varying delay and partially unknown transition probabilitiesrdquoSignal Processing vol 91 no 2 pp 277ndash289 2011
[39] J Tian Y Li J Zhao and S Zhong ldquoDelay-dependent stochasticstability criteria for Markovian jumping neural networks withmode-dependent time-varying delays and partially knowntransition ratesrdquo Applied Mathematics and Computation vol218 no 9 pp 5769ndash5781 2012
[40] Y Wei J Qiu H R Karimi and M Wang ldquoA new design 119867infin
filtering for contrinuous-time Markovian jump systems withtime-varying delay and partially accessible mode informationrdquoSignal Processing vol 93 pp 2392ndash2407 2013
[41] L Zhang E-K Boukas and J Lam ldquoAnalysis and synthesisof Markov jump linear systems with time-varying delays andpartially known transition probabilitiesrdquo IEEE Transactions onAutomatic Control vol 53 no 10 pp 2458ndash2464 2008
[42] L Zhang and E-K Boukas ldquoMode-dependent 119867infin
filteringfor discrete-time Markovian jump linear systems with partlyunknown transition probabilitiesrdquo Automatica vol 45 no 6pp 1462ndash1467 2009
[43] L Zhang and E-K Boukas ldquoMode-dependent 119867infin
controlfor discrete-time Markovian jump linear systems with partlyunknown transition probabilitiesrdquo International Journal ofRobust and Nonlinear Control vol 19 no 8 pp 868ndash883 2009
[44] Y Zhang Y He M Wu and J Zhang ldquoStabilization forMarkovian jump systems with partial information on transi-tion probability based on free-connection weighting matricesrdquoAutomatica vol 47 no 1 pp 79ndash84 2011
[45] G Zong D Yang L Hou and Q Wang ldquoRobust finite-time119867infincontrol for Markovian jump systems with partially known
transition probabilitiesrdquo Journal of the Franklin Institute vol350 no 6 pp 1562ndash1578 2013
[46] Y Ding H Zhu S Zhong and Y Zhang ldquo1198712minus 119871infin
filteringfor Markovian jump systems with time-varying delays andpartly unknown transition probabilitiesrdquo Communications inNonlinear Science and Numerical Simulation vol 17 no 7 pp3070ndash3081 2012
[47] Q Ma S Xu and Y Zou ldquoStability and synchronizationfor Markovian jump neural networks with partly unknowntransition probabilitiesrdquo Neurocomputing vol 74 pp 3403ndash3411 2011
[48] E Tian D Yue and G Wei ldquoRobust control for Markovianjump systems with partially known transition probabilities andnonlinearitiesrdquo Journal of the Franklin Institute vol 350 no 8pp 2069ndash2083 2013
[49] Y Guo and Z Wang ldquoStability of Markovian jump systemswith generally uncertain transition ratesrdquo Journal of the FranklinInstitute vol 350 no 9 pp 2826ndash2836 2013
[50] L Zhang ldquo119867infinestimation for discrete-time piecewise homoge-
neous Markov jump linear systemsrdquo Automatica vol 45 no 11pp 2570ndash2576 2009
[51] Z Wu H Su and J Chu ldquostate estimation for discreteMarkovian jumping neural networks with time delayrdquo Neuro-computing vol 73 pp 2247ndash2254 2010
[52] I R Petersen and A V Savkin Robust Kalman Filteringfor Signals and Systems with Large Uncertainties BirkhauserBoston Mass USA 1999
[53] B D O Anderson and J B Moore Optimal Filtering Prentice-Hall Upper Saddle River NJ USA 1979
[54] I R Petersen and D C McFarlane ldquoOptimal guaranteedcost control and filtering for uncertain linear systemsrdquo IEEETransactions on Automatic Control vol 39 no 9 pp 1971ndash19771994
[55] S Xu T Chen and J Lam ldquoRobust 119867infin filtering for uncertainMarkovian jump systems with mode-dependent time delaysrdquoIEEE Transactions on Automatic Control vol 48 no 5 pp 900ndash907 2003
[56] J H Kim D Oh and H Park ldquoGuaranteed cost and 119867infin
filtering for time-delay systemsrdquo in Proceeding of the AmericanControl Conference pp 4014ndash4018 IEEE Arlington Va USA2001
[57] L Xie ldquoOutput feedback119867infincontrol of systems with parameter
uncertaintyrdquo International Journal of Control vol 63 no 4 pp741ndash750 1996
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Abstract and Applied Analysis 7
Then a suitable robust guaranteed cost observer in theform of (11) has parameters as follows
1198701119894= 119875minus1
11198941198841119894 119870
2119894= 119875minus1
21198941198842119894
(47)
and 119869lowast is a robust guaranteed cost for system (13) with GUTRMatrix (4)
Proof Define
1198601
119894=
[[[[
[
119860119879
119891119894119875119894+ 119875119894119860119891119894+ 119876 + sum
119895isin119880119894
119896
119894119895(119875119895minus 119875119894) + sum
119895isin119880119894
119896
1
41205902
119894119895119879119894119895+ 119862119879
119891119862119891
119875119894119860119891119889119894
1198731
119860119879
119891119889119894119875119894
minus119876 0
lowast lowast 1198732
]]]]
]
(48)
1198602
119894=
[[[[
[
119860119879
119891119894119875119894+ 119875119894119860119891119894+ 119876 + sum
119895isin119880119894
119896
119894119895(119875119895minus 119875119894) + sum
119895isin119880119894
119896
1
41205902
119894119895119881119894119895119897+ 119862119879
119891119862119891
119875119894119860119891119889119894
1198721
119860119879
119891119889119894119875119894
minus119876 0
lowast lowast 1198722
]]]]
]
(49)
1198603
119894=
[[[[
[
119860119879
119891119894119875119894+ 119875119894119860119891119894+ 119876 + sum
119895isin119880119894
119896
119894119895(119875119895minus 119875119894) + sum
119895isin119880119894
119896
1
41205902
119894119895119882119894119895+ 119862119879
119891119862119891
119875119894119860119891119889119894
1198711
119860119879
119891119889119894119875119894
minus119876 0
lowast lowast 1198712
]]]]
]
lt 0 (50)
Then (16) is equivalent to
1198601
119894+ [
[
119875119894119872119891119894
0
0
]
]
119865119894[119873119891119894 1198731198911119894 0]
+ [119873119891119894 1198651198911119894 0]119879
119865119879
119894[
[
119875119894119872119891119894
0
0
]
]
119879
lt 0
(51)
Then (19) is equivalent to
1198602
119894+ [
[
119875119894119872119891119894
0
0
]
]
119865119894[119873119891119894 1198731198911119894 0]
+ [119873119891119894 1198651198911119894 0]119879
119865119879
119894[
[
119875119894119872119891119894
0
0
]
]
119879
lt 0
(52)
Then (21) is equivalent to
1198603
119894+ [
[
119875119894119872119891119894
0
0
]
]
119865119894[119873119891119894 1198731198911119894 0]
+ [119873119891119894 1198651198911119894 0]119879
119865119879
119894[
[
119875119894119872119891119894
0
0
]
]
119879
lt 0
(53)
By applying Lemma 24 in [57] (50) (51) and (52) hold forall uncertainties119865i satisfying119865
119879
119894119865119894lt 119868 if and only if there exist
positive scalars 120576119894 119894 = 1 2 119904 such that
1198601
119894+ 120576minus1
119894[
[
119875119894119872119891119894
0
0
]
]
[
[
119875119894119872119891119894
0
0
]
]
119879
+ 120576119894[119873119891119894 1198651198911119894 0]
119879
[119873119891119894 1198651198911119894 0] lt 0
1198602
119894+ 120576minus1
119894[
[
119875119894119872119891119894
0
0
]
]
[
[
119875119894119872119891119894
0
0
]
]
119879
+ 120576119894[119873119891119894 1198651198911119894 0]
119879
[119873119891119894 1198651198911119894 0] lt 0
1198603
119894+ 120576minus1
119894[
[
119875119894119872119891119894
0
0
]
]
[
[
119875119894119872119891119894
0
0
]
]
119879
+ 120576119894[119873119891119894 1198651198911119894 0]
119879
[119873119891119894 1198651198911119894 0] lt 0
(54)
Let 119875119894= diag(119875
1119894 1198752119894) and using (47) we can conclude from
Schur complement results that the above matrix inequalitiesare equivalent to the coupled LMIs (40) (43) and (45)It further follows from Theorem 10 that 119869
lowast is a robustguaranteed cost for system (13) with (4)
Remark 12 The solution of LMIs (39)ndash(45) parameterizesthe set of the proposed robust guaranteed cost observersThis parameterized representation can be used to design theguaranteed cost observer with some additional performanceconstraints By applying the methods in [14] the suboptimal
8 Abstract and Applied Analysis
guaranteed cost observer can be determined by solving acertain optimization problem This is the following theorem
Theorem 13 Consider system (13)with GUTRMatrix (4) andthe cost function (14) and suppose that the initial conditions 119903
0
and 1199091198910
are known if the following optimization problem
min1198761198751119894 1198752119894 120576119894 1198841119894 and 1198842119894
119869lowast
st LMIs (39) ndash (45)(55)
has a solution 119876 1198751119894 1198752119894 120576119894 1198841119894 and 119884
2119894 119894 = 1 2 119904
then the observer (11) is a suboptimal guaranteed costobserver for system (1) where 119869
lowast= 119909119879
1198910119864119879
1198911199030119875(1199030)1199091198910
+
tr(int0minus119889
1199091198910(119905)1199091198910(119905)119909119879
1198910119889119905119876)
Proof It follows from Theorem 11 that the observer (11)constructed in terms of the solution 119876 119875
1119894 1198752119894 120576119894 1198841119894 and
1198842119894 119894 = 1 2 119904 is a robust guaranteed cost observer By
noting that
int
0
minus119889
119909119879
1198910(119905) 119876119909
1198910(119905) 119889119905 = int
0
minus119889
tr (1199091198791198910
(119905) 1198761199091198910
(119905)) 119889119905
= tr(int0
minus119889
119909119879
1198910(119905) 1199091198910
(119905) 119889119905119876)
(56)
it follows that the suboptimal guaranteed cost observerproblem is turned into the minimization problem (55)
Remark 14 Theorem 13 gives the suboptimal guaranteed costobserver conditions of a class of linear Markovian jump-ing time-delay systems with generally incomplete transitionprobability and LMI constraints which can be easily solvedby the LMI toolbox in MATLAB
4 Numerical Example
In this section a numerical example is presented todemonstrate the effectiveness of the method mentioned inTheorem 11 Consider a 2-dimensional system (1) with 3Markovian switching modes In this numerical example thesingular system matrix is set as 119864 = [ 1 0
0 0] and the 3-mode
transition rate matrix is Λ = [minus32
2
15 21 minus36
] where Δ11 Δ31
isin
[minus015 015] Δ23 Δ33
isin [minus012 012] and Δ32
isin [minus01 01]The other system matrices are as follows
For mode 119894 = 1 there are
1198601= [
minus32 065
1 02] 119860
1198891= [
02 05
1 minus068]
1198621= [
[
12 065
minus65 19
minus021 minus18
]
]
1198621198891
= [
[
minus36 minus105
21 096
021 minus086
]
]
11987211
= [minus02
08] 119872
21= [
[
025
0875
minus2
]
]
11987311
= [minus12 31] 11987321
= [minus069 minus42]
(57)
For mode 119894 = 2 there are
1198602= [
minus1 6
2 minus36] 119860
1198892= [
minus31 minus16
3 075]
1198622= [
[
9 minus25
035 minus2
36 minus18
]
]
1198621198892
= [
[
089 minus6
minus12 09
minus24 6
]
]
11987212
= [23
minus4] 119872
22= [
[
075
minus36
25
]
]
11987312
= [minus72 minus6] 11987322
= [1 2]
(58)
For mode 119894 = 3 there are
1198603= [
minus106 29
minus03 36] 119860
1198893= [
minus56 minus12
minus3 45]
1198623= [
[
minus3 minus036
015 minus18
09 minus5
]
]
1198621198893
= [
[
minus165 5
minus12 265
minus098 minus56
]
]
11987213
= [minus82
minus03] 119872
23= [
[
minus052
25
minus36
]
]
11987313
= [105 minus5] 11987323
= [minus72 minus126]
(59)
Then we set the error state matrix 119871 = [ minus45 062 minus6
] and thepositive scalars in Theorem 11 are 120576
1= 02 120576
2= 015 120576
3=
032 According to the definitions of augmented statematricesin (12) we can easily obtain the following parameter matricesin Theorem 11 by MATLAB
11988411
= [minus84521006 00127
00127 84509001]
11988421
= [002 01291 01435
minus23520 minus04080 minus08106]
11988412
= [minus170991 269626
269626 minus246750]
11988422
= [minus200744 minus132941 529388
216893 180120 minus506693]
11988413
= [minus6751329 224456
224456 minus8976976]
11988423
= [minus136021 minus1461726 540500
minus34324 minus195125 minus103068]
1198751=[[[
[
63029 0 0 0
0 48620 0 0
0 0 22914 0
0 0 0 03169
]]]
]
Abstract and Applied Analysis 9
1198752=[[[
[
08914 0 0 0
0 12505 0 0
0 0 73629 0
0 0 0 30056
]]]
]
1198753=[[[
[
30265 0 0 0
0 02156 0 0
0 0 08965 0
0 0 0 10002
]]]
]
119876 =[[[
[
05000 0 0 0
0 05001 0 0
0 0 05001 0
0 0 0 05001
]]]
]
11987911
=[[[
[
34173214 minus8707765 0 0
minus8707765 4167216 0 0
0 0 22263598 minus3207456
0 0 minus3207456 12263101
]]]
]
11987923
=[[[
[
37753231 minus27999330 0 0
minus27999330 28107685 0 0
0 0 106907366 minus107432750
0 0 minus107432750 108555053
]]]
]
11987931
=[[[
[
9518504 minus5399245 0 0
minus5399245 8962029 0 0
0 0 14773012 minus2077540
0 0 minus2077540 14791256
]]]
]
11987932
=[[[
[
21617695 minus12094164 0 0
minus12094164 20371205 0 0
0 0 14786 minus09283
0 0 minus09283 14794
]]]
]
11987933
=[[[
[
14939313 minus8398780 0 0
minus8398780 14073689 0 0
0 0 1478123 minus1336452
0 0 minus1336452 2459347
]]]
]
11988111
=[[[
[
16650 0 0 0
0 16650 0 0
0 0 16650 0
0 0 0 16650
]]]
]
11988231
=[[[
[
15426 0 0 0
0 16650 0 0
0 0 16662 0
0 0 0 16650
]]]
]
11988232
=[[[
[
15428 0 0 0
0 16650 0 0
0 0 16622 0
0 0 0 16650
]]]
]
(60)
Therefore we can design a linear memoryless observer as(11) with the constant matrices
11987011
= 119875minus1
1111988411
= [minus13409860 00020
00026 17381530]
11987021
= 119875minus1
2111988421
= [00087 00563 00626
minus74219 minus12875 minus25579]
11987012
= 119875minus1
1211988412
= [minus191823 302475
215615 minus197321]
11987022
= 119875minus1
2211988422
= [minus27264 minus18056 71899
72163 59928 minus168583]
11987013
= 119875minus1
1311988413
= [minus2231402 74186
1041076 minus41637180]
11987023
= 119875minus1
2311988423
= [minus151724 minus1630481 602900
minus34317 minus195086 minus103047]
(61)
Finally the observer (11) with the above parameter matri-ces for this numerical example is a suboptimal guaranteedcost observer byTheorems 11 and 13
5 Conclusions
In this paper the robust guaranteed cost observer problemfor a class of uncertain descriptor time-delay systems withMarkovian jumping parameters and generally uncertaintransition rates is studied by using LMI method In thisGUTR singular model each transition rate can be com-pletely unknown or only its estimate value is known Theparameterrsquos uncertainty is time varying and is assumed tobe norm-bounded Memoryless guaranteed cost observersare designed in terms of a set of linear coupled matrixinequalities The suboptimal guaranteed cost observer isdesigned by solving a certain optimization problem Ourresults can be applicable to the general Markovian jumpsystems with bounded uncertain or partly uncertain TRmatrix
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This research is supported by the National NaturalScience Foundations of China (10701020 1097102260973048 61272077 60974025 60673101 and 60939003)NCET-08-0755 National 863 Plan Project (2008AA04Z401 2009AA043404) SDPW IMZQWH010016the Natural Science Foundation of Shandong Province (noZR2012FM006) the Scientific and Technological Projectof Shandong Province (no 2007GG3WZ04016) and theNatural Science Foundation of Guangxi Autonomous Region(no 2012GXNSFBA053003)
10 Abstract and Applied Analysis
References
[1] L Dai Singular Control Systems Springer Berlin Germany1989
[2] S Xu P V Dooren R Stefan and J Lam ldquoRobust stability andstabilization for singular systemswith state delay and parameteruncertaintyrdquo IEEE Transactions on Automatic Control vol 47no 7 pp 1122ndash1128 2002
[3] M Fu and G Duan ldquoRobust guaranteed cost observer foruncertain descriptor time-delay systems withMarkovian jump-ing parametersrdquo Acta Automatica Sinica vol 31 pp 479ndash4832005
[4] Y Q Xia E-K Boukas P Shi and J H Zhang ldquoStabilityand stabilization of continuous-time singular hybrid systemsrdquoAutomatica vol 45 no 6 pp 1504ndash1509 2009
[5] S Xu and J Lam Robust Control and Filtering of SingularSystems Springer Berlin Germany 2006
[6] Y-Y Cao and J Lam ldquoRobust 119867infin
control of uncertainMarkovian jump systems with time-delayrdquo IEEE Transactionson Automatic Control vol 45 no 1 pp 77ndash83 2000
[7] Y Li and Y Kao ldquoStability of stochastic reaction-diffusion sys-tems with Markovian switching and impulsive perturbationsrdquoMathematical Problems in Engineering vol 2012 Article ID429568 13 pages 2012
[8] X Mao and C Yuan Stochastic Differential Equations withMarkovian Switching Imperial College Press London UK2006
[9] E-K Boukas Stochastic Switching Systems Analysis and DesignBirkhauser Basel Switzerland 2005
[10] Y Kao C Wang and L Zhang ldquoDelay-dependent exponentialstability of impulsive markovian jumping Cohen-Grossbergneural networks with reaction-diffusion and mixed delaysrdquoNeural Processing Letters vol 38 no 3 pp 321ndash346 2013
[11] Y-G Kao J-F Guo C-H Wang and X-Q Sun ldquoDelay-dependent robust exponential stability of Markovian jump-ing reaction-diffusion Cohen-Grossberg neural networks withmixed delaysrdquo Journal of the Franklin Institute vol 349 no 6pp 1972ndash1988 2012
[12] Y Kao C Wang F Zha and H Cao ldquoStability in mean ofpartial variables for stochastic reactionmdashdiffusion systems withMarkovian switchingrdquo Journal of the Franklin Institute vol 351no 1 pp 500ndash512 2014
[13] N Kumaresan and P Balasubramaniam ldquoOptimal control forstochastic nonlinear singular system using neural networksrdquoComputers amp Mathematics with Applications vol 56 no 9 pp2145ndash2154 2008
[14] L Yu and J Chu ldquoAn LMI approach to guaranteed cost controlof linear uncertain time-delay systemsrdquoAutomatica vol 35 no6 pp 1155ndash1159 1999
[15] S Yin H Luo and S Ding ldquoReal-time implementation of fault-tolerant control systems with performance optimizationrdquo IEEETransactions on Industrial Electronics vol 64 no 5 pp 2402ndash2411 2014
[16] S Yin G Wang and H Karimi ldquoData-driven design of robustfault detection system for wind turbinesrdquoMechatronics 2013
[17] S Yin S X Ding A H A Sari and H Hao ldquoData-drivenmonitoring for stochastic systems and its application on batchprocessrdquo International Journal of Systems Science vol 44 no 7pp 1366ndash1376 2013
[18] S Yin S Ding A Haghani H Hao and P Zhang ldquoA com-parison study of basic datadriven fault diagnosis and process
monitoring methods on the benchmark Tennessee Eastmanprocessrdquo Journal of Process Control vol 22 no 9 pp 1567ndash15812012
[19] Y G Kao and C H Wang ldquoGlobal stability analysis forstochastic coupled reaction-diffusion systems on networksrdquoNonlinear Analysis Real World Applications vol 14 no 3 pp1457ndash1465 2013
[20] E K Boukas ldquoStabilization of stochastic singular nonlinearhybrid systemsrdquo Nonlinear Analysis Theory Methods amp Appli-cations vol 64 no 2 pp 217ndash228 2006
[21] E-K Boukas Control of Singular Systems with Random AbruptChanges Springer Berlin Germany 2008
[22] Y Ding H Zhu S Zhong and Y Zeng ldquoExponential mean-square stability of time-delay singular systems with Markovianswitching and nonlinear perturbationsrdquo Applied Mathematicsand Computation vol 219 no 4 pp 2350ndash2359 2012
[23] S Long S Zhong and Z Liu ldquoStochastic admissibility fora class of singular Markovian jump systems with mode-dependent time delaysrdquoAppliedMathematics and Computationvol 219 no 8 pp 4106ndash4117 2012
[24] Z Wu P Shi H Su and J Chu ldquoAsynchronous 1198682minus 119868infin
filtering for discrete-time stochastic Markov jump systems withrandomly occurred sensor nonlinearitiesrdquo Automatica vol 50no 1 pp 180ndash186 2014
[25] Z Wu P Shi H Su and J Chu ldquoStochastic synchronizationof Markovian jump neural networks with time-varying delayusing sampled-datardquo IEEE Transactions on Cybernetics vol 43no 6 pp 1796ndash1806 2013
[26] S Ma and E-K Boukas ldquoGuaranteed cost control of uncertaindiscrete-time singular Markov jump systems with indefinitequadratic costrdquo International Journal of Robust and NonlinearControl vol 21 no 9 pp 1031ndash1045 2011
[27] G Wang and Q Zhang ldquoRobust control of uncertain singularstochastic systems with Markovian switching via proportional-derivative state feedbackrdquo IET Control Theory amp Applicationsvol 6 no 8 pp 1089ndash1096 2012
[28] J Wang H Wang A Xue and R Lu ldquoDelay-dependent 119867infin
control for singular Markovian jump systems with time delayrdquoNonlinear Analysis Hybrid Systems vol 8 pp 1ndash12 2013
[29] L Wu X Su and P Shi ldquoSliding mode control with bounded1198712gain performance of Markovian jump singular time-delay
systemsrdquo Automatica vol 48 no 8 pp 1929ndash1933 2012[30] Z-GWu J H ParkH Su and J Chu ldquoStochastic stability anal-
ysis for discrete-time singular Markov jump systems with time-varying delay and piecewise-constant transition probabilitiesrdquoJournal of the Franklin Institute vol 349 no 9 pp 2889ndash29022012
[31] J Xiong J Lam H Gao and D W C Ho ldquoOn robust stabi-lization of Markovian jump systems with uncertain switchingprobabilitiesrdquo Automatica vol 41 no 5 pp 897ndash903 2005
[32] J Xiong and J Lam ldquoRobust 1198672control of Markovian jump
systems with uncertain switching probabilitiesrdquo InternationalJournal of Systems Science vol 40 no 3 pp 255ndash265 2009
[33] M Karan P Shi and C Y Kaya ldquoTransition probabilitybounds for the stochastic stability robustness of continuous-and discrete-timeMarkovian jump linear systemsrdquoAutomaticavol 42 no 12 pp 2159ndash2168 2006
[34] L Zhang and E-K Boukas ldquoStability and stabilization ofMarkovian jump linear systems with partly unknown transitionprobabilitiesrdquo Automatica vol 45 no 2 pp 463ndash468 2009
Abstract and Applied Analysis 11
[35] B Du J Lam Y Zou and Z Shu ldquoStability and stabilization forMarkovian jump time-delay systems with partially unknowntransition ratesrdquo IEEE Transactions on Circuits and Systems IRegular Papers vol 60 no 2 pp 341ndash351 2013
[36] L Zhang and J Lam ldquoNecessary and sufficient conditionsfor analysis and synthesis of Markov jump linear systemswith incomplete transition descriptionsrdquo IEEE Transactions onAutomatic Control vol 55 no 7 pp 1695ndash1701 2010
[37] Y Guo and F Zhu ldquoNew results on stability and stabilizationof Markovian jump systems with partly known transitionprobabilitiesrdquoMathematical Problems in Engineering vol 2012Article ID 869842 11 pages 2012
[38] J Lin S Fei and J Shen ldquoDelay-dependent 119867infin
filtering fordiscrete-time singular Markovian jump systems with time-varying delay and partially unknown transition probabilitiesrdquoSignal Processing vol 91 no 2 pp 277ndash289 2011
[39] J Tian Y Li J Zhao and S Zhong ldquoDelay-dependent stochasticstability criteria for Markovian jumping neural networks withmode-dependent time-varying delays and partially knowntransition ratesrdquo Applied Mathematics and Computation vol218 no 9 pp 5769ndash5781 2012
[40] Y Wei J Qiu H R Karimi and M Wang ldquoA new design 119867infin
filtering for contrinuous-time Markovian jump systems withtime-varying delay and partially accessible mode informationrdquoSignal Processing vol 93 pp 2392ndash2407 2013
[41] L Zhang E-K Boukas and J Lam ldquoAnalysis and synthesisof Markov jump linear systems with time-varying delays andpartially known transition probabilitiesrdquo IEEE Transactions onAutomatic Control vol 53 no 10 pp 2458ndash2464 2008
[42] L Zhang and E-K Boukas ldquoMode-dependent 119867infin
filteringfor discrete-time Markovian jump linear systems with partlyunknown transition probabilitiesrdquo Automatica vol 45 no 6pp 1462ndash1467 2009
[43] L Zhang and E-K Boukas ldquoMode-dependent 119867infin
controlfor discrete-time Markovian jump linear systems with partlyunknown transition probabilitiesrdquo International Journal ofRobust and Nonlinear Control vol 19 no 8 pp 868ndash883 2009
[44] Y Zhang Y He M Wu and J Zhang ldquoStabilization forMarkovian jump systems with partial information on transi-tion probability based on free-connection weighting matricesrdquoAutomatica vol 47 no 1 pp 79ndash84 2011
[45] G Zong D Yang L Hou and Q Wang ldquoRobust finite-time119867infincontrol for Markovian jump systems with partially known
transition probabilitiesrdquo Journal of the Franklin Institute vol350 no 6 pp 1562ndash1578 2013
[46] Y Ding H Zhu S Zhong and Y Zhang ldquo1198712minus 119871infin
filteringfor Markovian jump systems with time-varying delays andpartly unknown transition probabilitiesrdquo Communications inNonlinear Science and Numerical Simulation vol 17 no 7 pp3070ndash3081 2012
[47] Q Ma S Xu and Y Zou ldquoStability and synchronizationfor Markovian jump neural networks with partly unknowntransition probabilitiesrdquo Neurocomputing vol 74 pp 3403ndash3411 2011
[48] E Tian D Yue and G Wei ldquoRobust control for Markovianjump systems with partially known transition probabilities andnonlinearitiesrdquo Journal of the Franklin Institute vol 350 no 8pp 2069ndash2083 2013
[49] Y Guo and Z Wang ldquoStability of Markovian jump systemswith generally uncertain transition ratesrdquo Journal of the FranklinInstitute vol 350 no 9 pp 2826ndash2836 2013
[50] L Zhang ldquo119867infinestimation for discrete-time piecewise homoge-
neous Markov jump linear systemsrdquo Automatica vol 45 no 11pp 2570ndash2576 2009
[51] Z Wu H Su and J Chu ldquostate estimation for discreteMarkovian jumping neural networks with time delayrdquo Neuro-computing vol 73 pp 2247ndash2254 2010
[52] I R Petersen and A V Savkin Robust Kalman Filteringfor Signals and Systems with Large Uncertainties BirkhauserBoston Mass USA 1999
[53] B D O Anderson and J B Moore Optimal Filtering Prentice-Hall Upper Saddle River NJ USA 1979
[54] I R Petersen and D C McFarlane ldquoOptimal guaranteedcost control and filtering for uncertain linear systemsrdquo IEEETransactions on Automatic Control vol 39 no 9 pp 1971ndash19771994
[55] S Xu T Chen and J Lam ldquoRobust 119867infin filtering for uncertainMarkovian jump systems with mode-dependent time delaysrdquoIEEE Transactions on Automatic Control vol 48 no 5 pp 900ndash907 2003
[56] J H Kim D Oh and H Park ldquoGuaranteed cost and 119867infin
filtering for time-delay systemsrdquo in Proceeding of the AmericanControl Conference pp 4014ndash4018 IEEE Arlington Va USA2001
[57] L Xie ldquoOutput feedback119867infincontrol of systems with parameter
uncertaintyrdquo International Journal of Control vol 63 no 4 pp741ndash750 1996
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
8 Abstract and Applied Analysis
guaranteed cost observer can be determined by solving acertain optimization problem This is the following theorem
Theorem 13 Consider system (13)with GUTRMatrix (4) andthe cost function (14) and suppose that the initial conditions 119903
0
and 1199091198910
are known if the following optimization problem
min1198761198751119894 1198752119894 120576119894 1198841119894 and 1198842119894
119869lowast
st LMIs (39) ndash (45)(55)
has a solution 119876 1198751119894 1198752119894 120576119894 1198841119894 and 119884
2119894 119894 = 1 2 119904
then the observer (11) is a suboptimal guaranteed costobserver for system (1) where 119869
lowast= 119909119879
1198910119864119879
1198911199030119875(1199030)1199091198910
+
tr(int0minus119889
1199091198910(119905)1199091198910(119905)119909119879
1198910119889119905119876)
Proof It follows from Theorem 11 that the observer (11)constructed in terms of the solution 119876 119875
1119894 1198752119894 120576119894 1198841119894 and
1198842119894 119894 = 1 2 119904 is a robust guaranteed cost observer By
noting that
int
0
minus119889
119909119879
1198910(119905) 119876119909
1198910(119905) 119889119905 = int
0
minus119889
tr (1199091198791198910
(119905) 1198761199091198910
(119905)) 119889119905
= tr(int0
minus119889
119909119879
1198910(119905) 1199091198910
(119905) 119889119905119876)
(56)
it follows that the suboptimal guaranteed cost observerproblem is turned into the minimization problem (55)
Remark 14 Theorem 13 gives the suboptimal guaranteed costobserver conditions of a class of linear Markovian jump-ing time-delay systems with generally incomplete transitionprobability and LMI constraints which can be easily solvedby the LMI toolbox in MATLAB
4 Numerical Example
In this section a numerical example is presented todemonstrate the effectiveness of the method mentioned inTheorem 11 Consider a 2-dimensional system (1) with 3Markovian switching modes In this numerical example thesingular system matrix is set as 119864 = [ 1 0
0 0] and the 3-mode
transition rate matrix is Λ = [minus32
2
15 21 minus36
] where Δ11 Δ31
isin
[minus015 015] Δ23 Δ33
isin [minus012 012] and Δ32
isin [minus01 01]The other system matrices are as follows
For mode 119894 = 1 there are
1198601= [
minus32 065
1 02] 119860
1198891= [
02 05
1 minus068]
1198621= [
[
12 065
minus65 19
minus021 minus18
]
]
1198621198891
= [
[
minus36 minus105
21 096
021 minus086
]
]
11987211
= [minus02
08] 119872
21= [
[
025
0875
minus2
]
]
11987311
= [minus12 31] 11987321
= [minus069 minus42]
(57)
For mode 119894 = 2 there are
1198602= [
minus1 6
2 minus36] 119860
1198892= [
minus31 minus16
3 075]
1198622= [
[
9 minus25
035 minus2
36 minus18
]
]
1198621198892
= [
[
089 minus6
minus12 09
minus24 6
]
]
11987212
= [23
minus4] 119872
22= [
[
075
minus36
25
]
]
11987312
= [minus72 minus6] 11987322
= [1 2]
(58)
For mode 119894 = 3 there are
1198603= [
minus106 29
minus03 36] 119860
1198893= [
minus56 minus12
minus3 45]
1198623= [
[
minus3 minus036
015 minus18
09 minus5
]
]
1198621198893
= [
[
minus165 5
minus12 265
minus098 minus56
]
]
11987213
= [minus82
minus03] 119872
23= [
[
minus052
25
minus36
]
]
11987313
= [105 minus5] 11987323
= [minus72 minus126]
(59)
Then we set the error state matrix 119871 = [ minus45 062 minus6
] and thepositive scalars in Theorem 11 are 120576
1= 02 120576
2= 015 120576
3=
032 According to the definitions of augmented statematricesin (12) we can easily obtain the following parameter matricesin Theorem 11 by MATLAB
11988411
= [minus84521006 00127
00127 84509001]
11988421
= [002 01291 01435
minus23520 minus04080 minus08106]
11988412
= [minus170991 269626
269626 minus246750]
11988422
= [minus200744 minus132941 529388
216893 180120 minus506693]
11988413
= [minus6751329 224456
224456 minus8976976]
11988423
= [minus136021 minus1461726 540500
minus34324 minus195125 minus103068]
1198751=[[[
[
63029 0 0 0
0 48620 0 0
0 0 22914 0
0 0 0 03169
]]]
]
Abstract and Applied Analysis 9
1198752=[[[
[
08914 0 0 0
0 12505 0 0
0 0 73629 0
0 0 0 30056
]]]
]
1198753=[[[
[
30265 0 0 0
0 02156 0 0
0 0 08965 0
0 0 0 10002
]]]
]
119876 =[[[
[
05000 0 0 0
0 05001 0 0
0 0 05001 0
0 0 0 05001
]]]
]
11987911
=[[[
[
34173214 minus8707765 0 0
minus8707765 4167216 0 0
0 0 22263598 minus3207456
0 0 minus3207456 12263101
]]]
]
11987923
=[[[
[
37753231 minus27999330 0 0
minus27999330 28107685 0 0
0 0 106907366 minus107432750
0 0 minus107432750 108555053
]]]
]
11987931
=[[[
[
9518504 minus5399245 0 0
minus5399245 8962029 0 0
0 0 14773012 minus2077540
0 0 minus2077540 14791256
]]]
]
11987932
=[[[
[
21617695 minus12094164 0 0
minus12094164 20371205 0 0
0 0 14786 minus09283
0 0 minus09283 14794
]]]
]
11987933
=[[[
[
14939313 minus8398780 0 0
minus8398780 14073689 0 0
0 0 1478123 minus1336452
0 0 minus1336452 2459347
]]]
]
11988111
=[[[
[
16650 0 0 0
0 16650 0 0
0 0 16650 0
0 0 0 16650
]]]
]
11988231
=[[[
[
15426 0 0 0
0 16650 0 0
0 0 16662 0
0 0 0 16650
]]]
]
11988232
=[[[
[
15428 0 0 0
0 16650 0 0
0 0 16622 0
0 0 0 16650
]]]
]
(60)
Therefore we can design a linear memoryless observer as(11) with the constant matrices
11987011
= 119875minus1
1111988411
= [minus13409860 00020
00026 17381530]
11987021
= 119875minus1
2111988421
= [00087 00563 00626
minus74219 minus12875 minus25579]
11987012
= 119875minus1
1211988412
= [minus191823 302475
215615 minus197321]
11987022
= 119875minus1
2211988422
= [minus27264 minus18056 71899
72163 59928 minus168583]
11987013
= 119875minus1
1311988413
= [minus2231402 74186
1041076 minus41637180]
11987023
= 119875minus1
2311988423
= [minus151724 minus1630481 602900
minus34317 minus195086 minus103047]
(61)
Finally the observer (11) with the above parameter matri-ces for this numerical example is a suboptimal guaranteedcost observer byTheorems 11 and 13
5 Conclusions
In this paper the robust guaranteed cost observer problemfor a class of uncertain descriptor time-delay systems withMarkovian jumping parameters and generally uncertaintransition rates is studied by using LMI method In thisGUTR singular model each transition rate can be com-pletely unknown or only its estimate value is known Theparameterrsquos uncertainty is time varying and is assumed tobe norm-bounded Memoryless guaranteed cost observersare designed in terms of a set of linear coupled matrixinequalities The suboptimal guaranteed cost observer isdesigned by solving a certain optimization problem Ourresults can be applicable to the general Markovian jumpsystems with bounded uncertain or partly uncertain TRmatrix
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This research is supported by the National NaturalScience Foundations of China (10701020 1097102260973048 61272077 60974025 60673101 and 60939003)NCET-08-0755 National 863 Plan Project (2008AA04Z401 2009AA043404) SDPW IMZQWH010016the Natural Science Foundation of Shandong Province (noZR2012FM006) the Scientific and Technological Projectof Shandong Province (no 2007GG3WZ04016) and theNatural Science Foundation of Guangxi Autonomous Region(no 2012GXNSFBA053003)
10 Abstract and Applied Analysis
References
[1] L Dai Singular Control Systems Springer Berlin Germany1989
[2] S Xu P V Dooren R Stefan and J Lam ldquoRobust stability andstabilization for singular systemswith state delay and parameteruncertaintyrdquo IEEE Transactions on Automatic Control vol 47no 7 pp 1122ndash1128 2002
[3] M Fu and G Duan ldquoRobust guaranteed cost observer foruncertain descriptor time-delay systems withMarkovian jump-ing parametersrdquo Acta Automatica Sinica vol 31 pp 479ndash4832005
[4] Y Q Xia E-K Boukas P Shi and J H Zhang ldquoStabilityand stabilization of continuous-time singular hybrid systemsrdquoAutomatica vol 45 no 6 pp 1504ndash1509 2009
[5] S Xu and J Lam Robust Control and Filtering of SingularSystems Springer Berlin Germany 2006
[6] Y-Y Cao and J Lam ldquoRobust 119867infin
control of uncertainMarkovian jump systems with time-delayrdquo IEEE Transactionson Automatic Control vol 45 no 1 pp 77ndash83 2000
[7] Y Li and Y Kao ldquoStability of stochastic reaction-diffusion sys-tems with Markovian switching and impulsive perturbationsrdquoMathematical Problems in Engineering vol 2012 Article ID429568 13 pages 2012
[8] X Mao and C Yuan Stochastic Differential Equations withMarkovian Switching Imperial College Press London UK2006
[9] E-K Boukas Stochastic Switching Systems Analysis and DesignBirkhauser Basel Switzerland 2005
[10] Y Kao C Wang and L Zhang ldquoDelay-dependent exponentialstability of impulsive markovian jumping Cohen-Grossbergneural networks with reaction-diffusion and mixed delaysrdquoNeural Processing Letters vol 38 no 3 pp 321ndash346 2013
[11] Y-G Kao J-F Guo C-H Wang and X-Q Sun ldquoDelay-dependent robust exponential stability of Markovian jump-ing reaction-diffusion Cohen-Grossberg neural networks withmixed delaysrdquo Journal of the Franklin Institute vol 349 no 6pp 1972ndash1988 2012
[12] Y Kao C Wang F Zha and H Cao ldquoStability in mean ofpartial variables for stochastic reactionmdashdiffusion systems withMarkovian switchingrdquo Journal of the Franklin Institute vol 351no 1 pp 500ndash512 2014
[13] N Kumaresan and P Balasubramaniam ldquoOptimal control forstochastic nonlinear singular system using neural networksrdquoComputers amp Mathematics with Applications vol 56 no 9 pp2145ndash2154 2008
[14] L Yu and J Chu ldquoAn LMI approach to guaranteed cost controlof linear uncertain time-delay systemsrdquoAutomatica vol 35 no6 pp 1155ndash1159 1999
[15] S Yin H Luo and S Ding ldquoReal-time implementation of fault-tolerant control systems with performance optimizationrdquo IEEETransactions on Industrial Electronics vol 64 no 5 pp 2402ndash2411 2014
[16] S Yin G Wang and H Karimi ldquoData-driven design of robustfault detection system for wind turbinesrdquoMechatronics 2013
[17] S Yin S X Ding A H A Sari and H Hao ldquoData-drivenmonitoring for stochastic systems and its application on batchprocessrdquo International Journal of Systems Science vol 44 no 7pp 1366ndash1376 2013
[18] S Yin S Ding A Haghani H Hao and P Zhang ldquoA com-parison study of basic datadriven fault diagnosis and process
monitoring methods on the benchmark Tennessee Eastmanprocessrdquo Journal of Process Control vol 22 no 9 pp 1567ndash15812012
[19] Y G Kao and C H Wang ldquoGlobal stability analysis forstochastic coupled reaction-diffusion systems on networksrdquoNonlinear Analysis Real World Applications vol 14 no 3 pp1457ndash1465 2013
[20] E K Boukas ldquoStabilization of stochastic singular nonlinearhybrid systemsrdquo Nonlinear Analysis Theory Methods amp Appli-cations vol 64 no 2 pp 217ndash228 2006
[21] E-K Boukas Control of Singular Systems with Random AbruptChanges Springer Berlin Germany 2008
[22] Y Ding H Zhu S Zhong and Y Zeng ldquoExponential mean-square stability of time-delay singular systems with Markovianswitching and nonlinear perturbationsrdquo Applied Mathematicsand Computation vol 219 no 4 pp 2350ndash2359 2012
[23] S Long S Zhong and Z Liu ldquoStochastic admissibility fora class of singular Markovian jump systems with mode-dependent time delaysrdquoAppliedMathematics and Computationvol 219 no 8 pp 4106ndash4117 2012
[24] Z Wu P Shi H Su and J Chu ldquoAsynchronous 1198682minus 119868infin
filtering for discrete-time stochastic Markov jump systems withrandomly occurred sensor nonlinearitiesrdquo Automatica vol 50no 1 pp 180ndash186 2014
[25] Z Wu P Shi H Su and J Chu ldquoStochastic synchronizationof Markovian jump neural networks with time-varying delayusing sampled-datardquo IEEE Transactions on Cybernetics vol 43no 6 pp 1796ndash1806 2013
[26] S Ma and E-K Boukas ldquoGuaranteed cost control of uncertaindiscrete-time singular Markov jump systems with indefinitequadratic costrdquo International Journal of Robust and NonlinearControl vol 21 no 9 pp 1031ndash1045 2011
[27] G Wang and Q Zhang ldquoRobust control of uncertain singularstochastic systems with Markovian switching via proportional-derivative state feedbackrdquo IET Control Theory amp Applicationsvol 6 no 8 pp 1089ndash1096 2012
[28] J Wang H Wang A Xue and R Lu ldquoDelay-dependent 119867infin
control for singular Markovian jump systems with time delayrdquoNonlinear Analysis Hybrid Systems vol 8 pp 1ndash12 2013
[29] L Wu X Su and P Shi ldquoSliding mode control with bounded1198712gain performance of Markovian jump singular time-delay
systemsrdquo Automatica vol 48 no 8 pp 1929ndash1933 2012[30] Z-GWu J H ParkH Su and J Chu ldquoStochastic stability anal-
ysis for discrete-time singular Markov jump systems with time-varying delay and piecewise-constant transition probabilitiesrdquoJournal of the Franklin Institute vol 349 no 9 pp 2889ndash29022012
[31] J Xiong J Lam H Gao and D W C Ho ldquoOn robust stabi-lization of Markovian jump systems with uncertain switchingprobabilitiesrdquo Automatica vol 41 no 5 pp 897ndash903 2005
[32] J Xiong and J Lam ldquoRobust 1198672control of Markovian jump
systems with uncertain switching probabilitiesrdquo InternationalJournal of Systems Science vol 40 no 3 pp 255ndash265 2009
[33] M Karan P Shi and C Y Kaya ldquoTransition probabilitybounds for the stochastic stability robustness of continuous-and discrete-timeMarkovian jump linear systemsrdquoAutomaticavol 42 no 12 pp 2159ndash2168 2006
[34] L Zhang and E-K Boukas ldquoStability and stabilization ofMarkovian jump linear systems with partly unknown transitionprobabilitiesrdquo Automatica vol 45 no 2 pp 463ndash468 2009
Abstract and Applied Analysis 11
[35] B Du J Lam Y Zou and Z Shu ldquoStability and stabilization forMarkovian jump time-delay systems with partially unknowntransition ratesrdquo IEEE Transactions on Circuits and Systems IRegular Papers vol 60 no 2 pp 341ndash351 2013
[36] L Zhang and J Lam ldquoNecessary and sufficient conditionsfor analysis and synthesis of Markov jump linear systemswith incomplete transition descriptionsrdquo IEEE Transactions onAutomatic Control vol 55 no 7 pp 1695ndash1701 2010
[37] Y Guo and F Zhu ldquoNew results on stability and stabilizationof Markovian jump systems with partly known transitionprobabilitiesrdquoMathematical Problems in Engineering vol 2012Article ID 869842 11 pages 2012
[38] J Lin S Fei and J Shen ldquoDelay-dependent 119867infin
filtering fordiscrete-time singular Markovian jump systems with time-varying delay and partially unknown transition probabilitiesrdquoSignal Processing vol 91 no 2 pp 277ndash289 2011
[39] J Tian Y Li J Zhao and S Zhong ldquoDelay-dependent stochasticstability criteria for Markovian jumping neural networks withmode-dependent time-varying delays and partially knowntransition ratesrdquo Applied Mathematics and Computation vol218 no 9 pp 5769ndash5781 2012
[40] Y Wei J Qiu H R Karimi and M Wang ldquoA new design 119867infin
filtering for contrinuous-time Markovian jump systems withtime-varying delay and partially accessible mode informationrdquoSignal Processing vol 93 pp 2392ndash2407 2013
[41] L Zhang E-K Boukas and J Lam ldquoAnalysis and synthesisof Markov jump linear systems with time-varying delays andpartially known transition probabilitiesrdquo IEEE Transactions onAutomatic Control vol 53 no 10 pp 2458ndash2464 2008
[42] L Zhang and E-K Boukas ldquoMode-dependent 119867infin
filteringfor discrete-time Markovian jump linear systems with partlyunknown transition probabilitiesrdquo Automatica vol 45 no 6pp 1462ndash1467 2009
[43] L Zhang and E-K Boukas ldquoMode-dependent 119867infin
controlfor discrete-time Markovian jump linear systems with partlyunknown transition probabilitiesrdquo International Journal ofRobust and Nonlinear Control vol 19 no 8 pp 868ndash883 2009
[44] Y Zhang Y He M Wu and J Zhang ldquoStabilization forMarkovian jump systems with partial information on transi-tion probability based on free-connection weighting matricesrdquoAutomatica vol 47 no 1 pp 79ndash84 2011
[45] G Zong D Yang L Hou and Q Wang ldquoRobust finite-time119867infincontrol for Markovian jump systems with partially known
transition probabilitiesrdquo Journal of the Franklin Institute vol350 no 6 pp 1562ndash1578 2013
[46] Y Ding H Zhu S Zhong and Y Zhang ldquo1198712minus 119871infin
filteringfor Markovian jump systems with time-varying delays andpartly unknown transition probabilitiesrdquo Communications inNonlinear Science and Numerical Simulation vol 17 no 7 pp3070ndash3081 2012
[47] Q Ma S Xu and Y Zou ldquoStability and synchronizationfor Markovian jump neural networks with partly unknowntransition probabilitiesrdquo Neurocomputing vol 74 pp 3403ndash3411 2011
[48] E Tian D Yue and G Wei ldquoRobust control for Markovianjump systems with partially known transition probabilities andnonlinearitiesrdquo Journal of the Franklin Institute vol 350 no 8pp 2069ndash2083 2013
[49] Y Guo and Z Wang ldquoStability of Markovian jump systemswith generally uncertain transition ratesrdquo Journal of the FranklinInstitute vol 350 no 9 pp 2826ndash2836 2013
[50] L Zhang ldquo119867infinestimation for discrete-time piecewise homoge-
neous Markov jump linear systemsrdquo Automatica vol 45 no 11pp 2570ndash2576 2009
[51] Z Wu H Su and J Chu ldquostate estimation for discreteMarkovian jumping neural networks with time delayrdquo Neuro-computing vol 73 pp 2247ndash2254 2010
[52] I R Petersen and A V Savkin Robust Kalman Filteringfor Signals and Systems with Large Uncertainties BirkhauserBoston Mass USA 1999
[53] B D O Anderson and J B Moore Optimal Filtering Prentice-Hall Upper Saddle River NJ USA 1979
[54] I R Petersen and D C McFarlane ldquoOptimal guaranteedcost control and filtering for uncertain linear systemsrdquo IEEETransactions on Automatic Control vol 39 no 9 pp 1971ndash19771994
[55] S Xu T Chen and J Lam ldquoRobust 119867infin filtering for uncertainMarkovian jump systems with mode-dependent time delaysrdquoIEEE Transactions on Automatic Control vol 48 no 5 pp 900ndash907 2003
[56] J H Kim D Oh and H Park ldquoGuaranteed cost and 119867infin
filtering for time-delay systemsrdquo in Proceeding of the AmericanControl Conference pp 4014ndash4018 IEEE Arlington Va USA2001
[57] L Xie ldquoOutput feedback119867infincontrol of systems with parameter
uncertaintyrdquo International Journal of Control vol 63 no 4 pp741ndash750 1996
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Abstract and Applied Analysis 9
1198752=[[[
[
08914 0 0 0
0 12505 0 0
0 0 73629 0
0 0 0 30056
]]]
]
1198753=[[[
[
30265 0 0 0
0 02156 0 0
0 0 08965 0
0 0 0 10002
]]]
]
119876 =[[[
[
05000 0 0 0
0 05001 0 0
0 0 05001 0
0 0 0 05001
]]]
]
11987911
=[[[
[
34173214 minus8707765 0 0
minus8707765 4167216 0 0
0 0 22263598 minus3207456
0 0 minus3207456 12263101
]]]
]
11987923
=[[[
[
37753231 minus27999330 0 0
minus27999330 28107685 0 0
0 0 106907366 minus107432750
0 0 minus107432750 108555053
]]]
]
11987931
=[[[
[
9518504 minus5399245 0 0
minus5399245 8962029 0 0
0 0 14773012 minus2077540
0 0 minus2077540 14791256
]]]
]
11987932
=[[[
[
21617695 minus12094164 0 0
minus12094164 20371205 0 0
0 0 14786 minus09283
0 0 minus09283 14794
]]]
]
11987933
=[[[
[
14939313 minus8398780 0 0
minus8398780 14073689 0 0
0 0 1478123 minus1336452
0 0 minus1336452 2459347
]]]
]
11988111
=[[[
[
16650 0 0 0
0 16650 0 0
0 0 16650 0
0 0 0 16650
]]]
]
11988231
=[[[
[
15426 0 0 0
0 16650 0 0
0 0 16662 0
0 0 0 16650
]]]
]
11988232
=[[[
[
15428 0 0 0
0 16650 0 0
0 0 16622 0
0 0 0 16650
]]]
]
(60)
Therefore we can design a linear memoryless observer as(11) with the constant matrices
11987011
= 119875minus1
1111988411
= [minus13409860 00020
00026 17381530]
11987021
= 119875minus1
2111988421
= [00087 00563 00626
minus74219 minus12875 minus25579]
11987012
= 119875minus1
1211988412
= [minus191823 302475
215615 minus197321]
11987022
= 119875minus1
2211988422
= [minus27264 minus18056 71899
72163 59928 minus168583]
11987013
= 119875minus1
1311988413
= [minus2231402 74186
1041076 minus41637180]
11987023
= 119875minus1
2311988423
= [minus151724 minus1630481 602900
minus34317 minus195086 minus103047]
(61)
Finally the observer (11) with the above parameter matri-ces for this numerical example is a suboptimal guaranteedcost observer byTheorems 11 and 13
5 Conclusions
In this paper the robust guaranteed cost observer problemfor a class of uncertain descriptor time-delay systems withMarkovian jumping parameters and generally uncertaintransition rates is studied by using LMI method In thisGUTR singular model each transition rate can be com-pletely unknown or only its estimate value is known Theparameterrsquos uncertainty is time varying and is assumed tobe norm-bounded Memoryless guaranteed cost observersare designed in terms of a set of linear coupled matrixinequalities The suboptimal guaranteed cost observer isdesigned by solving a certain optimization problem Ourresults can be applicable to the general Markovian jumpsystems with bounded uncertain or partly uncertain TRmatrix
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This research is supported by the National NaturalScience Foundations of China (10701020 1097102260973048 61272077 60974025 60673101 and 60939003)NCET-08-0755 National 863 Plan Project (2008AA04Z401 2009AA043404) SDPW IMZQWH010016the Natural Science Foundation of Shandong Province (noZR2012FM006) the Scientific and Technological Projectof Shandong Province (no 2007GG3WZ04016) and theNatural Science Foundation of Guangxi Autonomous Region(no 2012GXNSFBA053003)
10 Abstract and Applied Analysis
References
[1] L Dai Singular Control Systems Springer Berlin Germany1989
[2] S Xu P V Dooren R Stefan and J Lam ldquoRobust stability andstabilization for singular systemswith state delay and parameteruncertaintyrdquo IEEE Transactions on Automatic Control vol 47no 7 pp 1122ndash1128 2002
[3] M Fu and G Duan ldquoRobust guaranteed cost observer foruncertain descriptor time-delay systems withMarkovian jump-ing parametersrdquo Acta Automatica Sinica vol 31 pp 479ndash4832005
[4] Y Q Xia E-K Boukas P Shi and J H Zhang ldquoStabilityand stabilization of continuous-time singular hybrid systemsrdquoAutomatica vol 45 no 6 pp 1504ndash1509 2009
[5] S Xu and J Lam Robust Control and Filtering of SingularSystems Springer Berlin Germany 2006
[6] Y-Y Cao and J Lam ldquoRobust 119867infin
control of uncertainMarkovian jump systems with time-delayrdquo IEEE Transactionson Automatic Control vol 45 no 1 pp 77ndash83 2000
[7] Y Li and Y Kao ldquoStability of stochastic reaction-diffusion sys-tems with Markovian switching and impulsive perturbationsrdquoMathematical Problems in Engineering vol 2012 Article ID429568 13 pages 2012
[8] X Mao and C Yuan Stochastic Differential Equations withMarkovian Switching Imperial College Press London UK2006
[9] E-K Boukas Stochastic Switching Systems Analysis and DesignBirkhauser Basel Switzerland 2005
[10] Y Kao C Wang and L Zhang ldquoDelay-dependent exponentialstability of impulsive markovian jumping Cohen-Grossbergneural networks with reaction-diffusion and mixed delaysrdquoNeural Processing Letters vol 38 no 3 pp 321ndash346 2013
[11] Y-G Kao J-F Guo C-H Wang and X-Q Sun ldquoDelay-dependent robust exponential stability of Markovian jump-ing reaction-diffusion Cohen-Grossberg neural networks withmixed delaysrdquo Journal of the Franklin Institute vol 349 no 6pp 1972ndash1988 2012
[12] Y Kao C Wang F Zha and H Cao ldquoStability in mean ofpartial variables for stochastic reactionmdashdiffusion systems withMarkovian switchingrdquo Journal of the Franklin Institute vol 351no 1 pp 500ndash512 2014
[13] N Kumaresan and P Balasubramaniam ldquoOptimal control forstochastic nonlinear singular system using neural networksrdquoComputers amp Mathematics with Applications vol 56 no 9 pp2145ndash2154 2008
[14] L Yu and J Chu ldquoAn LMI approach to guaranteed cost controlof linear uncertain time-delay systemsrdquoAutomatica vol 35 no6 pp 1155ndash1159 1999
[15] S Yin H Luo and S Ding ldquoReal-time implementation of fault-tolerant control systems with performance optimizationrdquo IEEETransactions on Industrial Electronics vol 64 no 5 pp 2402ndash2411 2014
[16] S Yin G Wang and H Karimi ldquoData-driven design of robustfault detection system for wind turbinesrdquoMechatronics 2013
[17] S Yin S X Ding A H A Sari and H Hao ldquoData-drivenmonitoring for stochastic systems and its application on batchprocessrdquo International Journal of Systems Science vol 44 no 7pp 1366ndash1376 2013
[18] S Yin S Ding A Haghani H Hao and P Zhang ldquoA com-parison study of basic datadriven fault diagnosis and process
monitoring methods on the benchmark Tennessee Eastmanprocessrdquo Journal of Process Control vol 22 no 9 pp 1567ndash15812012
[19] Y G Kao and C H Wang ldquoGlobal stability analysis forstochastic coupled reaction-diffusion systems on networksrdquoNonlinear Analysis Real World Applications vol 14 no 3 pp1457ndash1465 2013
[20] E K Boukas ldquoStabilization of stochastic singular nonlinearhybrid systemsrdquo Nonlinear Analysis Theory Methods amp Appli-cations vol 64 no 2 pp 217ndash228 2006
[21] E-K Boukas Control of Singular Systems with Random AbruptChanges Springer Berlin Germany 2008
[22] Y Ding H Zhu S Zhong and Y Zeng ldquoExponential mean-square stability of time-delay singular systems with Markovianswitching and nonlinear perturbationsrdquo Applied Mathematicsand Computation vol 219 no 4 pp 2350ndash2359 2012
[23] S Long S Zhong and Z Liu ldquoStochastic admissibility fora class of singular Markovian jump systems with mode-dependent time delaysrdquoAppliedMathematics and Computationvol 219 no 8 pp 4106ndash4117 2012
[24] Z Wu P Shi H Su and J Chu ldquoAsynchronous 1198682minus 119868infin
filtering for discrete-time stochastic Markov jump systems withrandomly occurred sensor nonlinearitiesrdquo Automatica vol 50no 1 pp 180ndash186 2014
[25] Z Wu P Shi H Su and J Chu ldquoStochastic synchronizationof Markovian jump neural networks with time-varying delayusing sampled-datardquo IEEE Transactions on Cybernetics vol 43no 6 pp 1796ndash1806 2013
[26] S Ma and E-K Boukas ldquoGuaranteed cost control of uncertaindiscrete-time singular Markov jump systems with indefinitequadratic costrdquo International Journal of Robust and NonlinearControl vol 21 no 9 pp 1031ndash1045 2011
[27] G Wang and Q Zhang ldquoRobust control of uncertain singularstochastic systems with Markovian switching via proportional-derivative state feedbackrdquo IET Control Theory amp Applicationsvol 6 no 8 pp 1089ndash1096 2012
[28] J Wang H Wang A Xue and R Lu ldquoDelay-dependent 119867infin
control for singular Markovian jump systems with time delayrdquoNonlinear Analysis Hybrid Systems vol 8 pp 1ndash12 2013
[29] L Wu X Su and P Shi ldquoSliding mode control with bounded1198712gain performance of Markovian jump singular time-delay
systemsrdquo Automatica vol 48 no 8 pp 1929ndash1933 2012[30] Z-GWu J H ParkH Su and J Chu ldquoStochastic stability anal-
ysis for discrete-time singular Markov jump systems with time-varying delay and piecewise-constant transition probabilitiesrdquoJournal of the Franklin Institute vol 349 no 9 pp 2889ndash29022012
[31] J Xiong J Lam H Gao and D W C Ho ldquoOn robust stabi-lization of Markovian jump systems with uncertain switchingprobabilitiesrdquo Automatica vol 41 no 5 pp 897ndash903 2005
[32] J Xiong and J Lam ldquoRobust 1198672control of Markovian jump
systems with uncertain switching probabilitiesrdquo InternationalJournal of Systems Science vol 40 no 3 pp 255ndash265 2009
[33] M Karan P Shi and C Y Kaya ldquoTransition probabilitybounds for the stochastic stability robustness of continuous-and discrete-timeMarkovian jump linear systemsrdquoAutomaticavol 42 no 12 pp 2159ndash2168 2006
[34] L Zhang and E-K Boukas ldquoStability and stabilization ofMarkovian jump linear systems with partly unknown transitionprobabilitiesrdquo Automatica vol 45 no 2 pp 463ndash468 2009
Abstract and Applied Analysis 11
[35] B Du J Lam Y Zou and Z Shu ldquoStability and stabilization forMarkovian jump time-delay systems with partially unknowntransition ratesrdquo IEEE Transactions on Circuits and Systems IRegular Papers vol 60 no 2 pp 341ndash351 2013
[36] L Zhang and J Lam ldquoNecessary and sufficient conditionsfor analysis and synthesis of Markov jump linear systemswith incomplete transition descriptionsrdquo IEEE Transactions onAutomatic Control vol 55 no 7 pp 1695ndash1701 2010
[37] Y Guo and F Zhu ldquoNew results on stability and stabilizationof Markovian jump systems with partly known transitionprobabilitiesrdquoMathematical Problems in Engineering vol 2012Article ID 869842 11 pages 2012
[38] J Lin S Fei and J Shen ldquoDelay-dependent 119867infin
filtering fordiscrete-time singular Markovian jump systems with time-varying delay and partially unknown transition probabilitiesrdquoSignal Processing vol 91 no 2 pp 277ndash289 2011
[39] J Tian Y Li J Zhao and S Zhong ldquoDelay-dependent stochasticstability criteria for Markovian jumping neural networks withmode-dependent time-varying delays and partially knowntransition ratesrdquo Applied Mathematics and Computation vol218 no 9 pp 5769ndash5781 2012
[40] Y Wei J Qiu H R Karimi and M Wang ldquoA new design 119867infin
filtering for contrinuous-time Markovian jump systems withtime-varying delay and partially accessible mode informationrdquoSignal Processing vol 93 pp 2392ndash2407 2013
[41] L Zhang E-K Boukas and J Lam ldquoAnalysis and synthesisof Markov jump linear systems with time-varying delays andpartially known transition probabilitiesrdquo IEEE Transactions onAutomatic Control vol 53 no 10 pp 2458ndash2464 2008
[42] L Zhang and E-K Boukas ldquoMode-dependent 119867infin
filteringfor discrete-time Markovian jump linear systems with partlyunknown transition probabilitiesrdquo Automatica vol 45 no 6pp 1462ndash1467 2009
[43] L Zhang and E-K Boukas ldquoMode-dependent 119867infin
controlfor discrete-time Markovian jump linear systems with partlyunknown transition probabilitiesrdquo International Journal ofRobust and Nonlinear Control vol 19 no 8 pp 868ndash883 2009
[44] Y Zhang Y He M Wu and J Zhang ldquoStabilization forMarkovian jump systems with partial information on transi-tion probability based on free-connection weighting matricesrdquoAutomatica vol 47 no 1 pp 79ndash84 2011
[45] G Zong D Yang L Hou and Q Wang ldquoRobust finite-time119867infincontrol for Markovian jump systems with partially known
transition probabilitiesrdquo Journal of the Franklin Institute vol350 no 6 pp 1562ndash1578 2013
[46] Y Ding H Zhu S Zhong and Y Zhang ldquo1198712minus 119871infin
filteringfor Markovian jump systems with time-varying delays andpartly unknown transition probabilitiesrdquo Communications inNonlinear Science and Numerical Simulation vol 17 no 7 pp3070ndash3081 2012
[47] Q Ma S Xu and Y Zou ldquoStability and synchronizationfor Markovian jump neural networks with partly unknowntransition probabilitiesrdquo Neurocomputing vol 74 pp 3403ndash3411 2011
[48] E Tian D Yue and G Wei ldquoRobust control for Markovianjump systems with partially known transition probabilities andnonlinearitiesrdquo Journal of the Franklin Institute vol 350 no 8pp 2069ndash2083 2013
[49] Y Guo and Z Wang ldquoStability of Markovian jump systemswith generally uncertain transition ratesrdquo Journal of the FranklinInstitute vol 350 no 9 pp 2826ndash2836 2013
[50] L Zhang ldquo119867infinestimation for discrete-time piecewise homoge-
neous Markov jump linear systemsrdquo Automatica vol 45 no 11pp 2570ndash2576 2009
[51] Z Wu H Su and J Chu ldquostate estimation for discreteMarkovian jumping neural networks with time delayrdquo Neuro-computing vol 73 pp 2247ndash2254 2010
[52] I R Petersen and A V Savkin Robust Kalman Filteringfor Signals and Systems with Large Uncertainties BirkhauserBoston Mass USA 1999
[53] B D O Anderson and J B Moore Optimal Filtering Prentice-Hall Upper Saddle River NJ USA 1979
[54] I R Petersen and D C McFarlane ldquoOptimal guaranteedcost control and filtering for uncertain linear systemsrdquo IEEETransactions on Automatic Control vol 39 no 9 pp 1971ndash19771994
[55] S Xu T Chen and J Lam ldquoRobust 119867infin filtering for uncertainMarkovian jump systems with mode-dependent time delaysrdquoIEEE Transactions on Automatic Control vol 48 no 5 pp 900ndash907 2003
[56] J H Kim D Oh and H Park ldquoGuaranteed cost and 119867infin
filtering for time-delay systemsrdquo in Proceeding of the AmericanControl Conference pp 4014ndash4018 IEEE Arlington Va USA2001
[57] L Xie ldquoOutput feedback119867infincontrol of systems with parameter
uncertaintyrdquo International Journal of Control vol 63 no 4 pp741ndash750 1996
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
10 Abstract and Applied Analysis
References
[1] L Dai Singular Control Systems Springer Berlin Germany1989
[2] S Xu P V Dooren R Stefan and J Lam ldquoRobust stability andstabilization for singular systemswith state delay and parameteruncertaintyrdquo IEEE Transactions on Automatic Control vol 47no 7 pp 1122ndash1128 2002
[3] M Fu and G Duan ldquoRobust guaranteed cost observer foruncertain descriptor time-delay systems withMarkovian jump-ing parametersrdquo Acta Automatica Sinica vol 31 pp 479ndash4832005
[4] Y Q Xia E-K Boukas P Shi and J H Zhang ldquoStabilityand stabilization of continuous-time singular hybrid systemsrdquoAutomatica vol 45 no 6 pp 1504ndash1509 2009
[5] S Xu and J Lam Robust Control and Filtering of SingularSystems Springer Berlin Germany 2006
[6] Y-Y Cao and J Lam ldquoRobust 119867infin
control of uncertainMarkovian jump systems with time-delayrdquo IEEE Transactionson Automatic Control vol 45 no 1 pp 77ndash83 2000
[7] Y Li and Y Kao ldquoStability of stochastic reaction-diffusion sys-tems with Markovian switching and impulsive perturbationsrdquoMathematical Problems in Engineering vol 2012 Article ID429568 13 pages 2012
[8] X Mao and C Yuan Stochastic Differential Equations withMarkovian Switching Imperial College Press London UK2006
[9] E-K Boukas Stochastic Switching Systems Analysis and DesignBirkhauser Basel Switzerland 2005
[10] Y Kao C Wang and L Zhang ldquoDelay-dependent exponentialstability of impulsive markovian jumping Cohen-Grossbergneural networks with reaction-diffusion and mixed delaysrdquoNeural Processing Letters vol 38 no 3 pp 321ndash346 2013
[11] Y-G Kao J-F Guo C-H Wang and X-Q Sun ldquoDelay-dependent robust exponential stability of Markovian jump-ing reaction-diffusion Cohen-Grossberg neural networks withmixed delaysrdquo Journal of the Franklin Institute vol 349 no 6pp 1972ndash1988 2012
[12] Y Kao C Wang F Zha and H Cao ldquoStability in mean ofpartial variables for stochastic reactionmdashdiffusion systems withMarkovian switchingrdquo Journal of the Franklin Institute vol 351no 1 pp 500ndash512 2014
[13] N Kumaresan and P Balasubramaniam ldquoOptimal control forstochastic nonlinear singular system using neural networksrdquoComputers amp Mathematics with Applications vol 56 no 9 pp2145ndash2154 2008
[14] L Yu and J Chu ldquoAn LMI approach to guaranteed cost controlof linear uncertain time-delay systemsrdquoAutomatica vol 35 no6 pp 1155ndash1159 1999
[15] S Yin H Luo and S Ding ldquoReal-time implementation of fault-tolerant control systems with performance optimizationrdquo IEEETransactions on Industrial Electronics vol 64 no 5 pp 2402ndash2411 2014
[16] S Yin G Wang and H Karimi ldquoData-driven design of robustfault detection system for wind turbinesrdquoMechatronics 2013
[17] S Yin S X Ding A H A Sari and H Hao ldquoData-drivenmonitoring for stochastic systems and its application on batchprocessrdquo International Journal of Systems Science vol 44 no 7pp 1366ndash1376 2013
[18] S Yin S Ding A Haghani H Hao and P Zhang ldquoA com-parison study of basic datadriven fault diagnosis and process
monitoring methods on the benchmark Tennessee Eastmanprocessrdquo Journal of Process Control vol 22 no 9 pp 1567ndash15812012
[19] Y G Kao and C H Wang ldquoGlobal stability analysis forstochastic coupled reaction-diffusion systems on networksrdquoNonlinear Analysis Real World Applications vol 14 no 3 pp1457ndash1465 2013
[20] E K Boukas ldquoStabilization of stochastic singular nonlinearhybrid systemsrdquo Nonlinear Analysis Theory Methods amp Appli-cations vol 64 no 2 pp 217ndash228 2006
[21] E-K Boukas Control of Singular Systems with Random AbruptChanges Springer Berlin Germany 2008
[22] Y Ding H Zhu S Zhong and Y Zeng ldquoExponential mean-square stability of time-delay singular systems with Markovianswitching and nonlinear perturbationsrdquo Applied Mathematicsand Computation vol 219 no 4 pp 2350ndash2359 2012
[23] S Long S Zhong and Z Liu ldquoStochastic admissibility fora class of singular Markovian jump systems with mode-dependent time delaysrdquoAppliedMathematics and Computationvol 219 no 8 pp 4106ndash4117 2012
[24] Z Wu P Shi H Su and J Chu ldquoAsynchronous 1198682minus 119868infin
filtering for discrete-time stochastic Markov jump systems withrandomly occurred sensor nonlinearitiesrdquo Automatica vol 50no 1 pp 180ndash186 2014
[25] Z Wu P Shi H Su and J Chu ldquoStochastic synchronizationof Markovian jump neural networks with time-varying delayusing sampled-datardquo IEEE Transactions on Cybernetics vol 43no 6 pp 1796ndash1806 2013
[26] S Ma and E-K Boukas ldquoGuaranteed cost control of uncertaindiscrete-time singular Markov jump systems with indefinitequadratic costrdquo International Journal of Robust and NonlinearControl vol 21 no 9 pp 1031ndash1045 2011
[27] G Wang and Q Zhang ldquoRobust control of uncertain singularstochastic systems with Markovian switching via proportional-derivative state feedbackrdquo IET Control Theory amp Applicationsvol 6 no 8 pp 1089ndash1096 2012
[28] J Wang H Wang A Xue and R Lu ldquoDelay-dependent 119867infin
control for singular Markovian jump systems with time delayrdquoNonlinear Analysis Hybrid Systems vol 8 pp 1ndash12 2013
[29] L Wu X Su and P Shi ldquoSliding mode control with bounded1198712gain performance of Markovian jump singular time-delay
systemsrdquo Automatica vol 48 no 8 pp 1929ndash1933 2012[30] Z-GWu J H ParkH Su and J Chu ldquoStochastic stability anal-
ysis for discrete-time singular Markov jump systems with time-varying delay and piecewise-constant transition probabilitiesrdquoJournal of the Franklin Institute vol 349 no 9 pp 2889ndash29022012
[31] J Xiong J Lam H Gao and D W C Ho ldquoOn robust stabi-lization of Markovian jump systems with uncertain switchingprobabilitiesrdquo Automatica vol 41 no 5 pp 897ndash903 2005
[32] J Xiong and J Lam ldquoRobust 1198672control of Markovian jump
systems with uncertain switching probabilitiesrdquo InternationalJournal of Systems Science vol 40 no 3 pp 255ndash265 2009
[33] M Karan P Shi and C Y Kaya ldquoTransition probabilitybounds for the stochastic stability robustness of continuous-and discrete-timeMarkovian jump linear systemsrdquoAutomaticavol 42 no 12 pp 2159ndash2168 2006
[34] L Zhang and E-K Boukas ldquoStability and stabilization ofMarkovian jump linear systems with partly unknown transitionprobabilitiesrdquo Automatica vol 45 no 2 pp 463ndash468 2009
Abstract and Applied Analysis 11
[35] B Du J Lam Y Zou and Z Shu ldquoStability and stabilization forMarkovian jump time-delay systems with partially unknowntransition ratesrdquo IEEE Transactions on Circuits and Systems IRegular Papers vol 60 no 2 pp 341ndash351 2013
[36] L Zhang and J Lam ldquoNecessary and sufficient conditionsfor analysis and synthesis of Markov jump linear systemswith incomplete transition descriptionsrdquo IEEE Transactions onAutomatic Control vol 55 no 7 pp 1695ndash1701 2010
[37] Y Guo and F Zhu ldquoNew results on stability and stabilizationof Markovian jump systems with partly known transitionprobabilitiesrdquoMathematical Problems in Engineering vol 2012Article ID 869842 11 pages 2012
[38] J Lin S Fei and J Shen ldquoDelay-dependent 119867infin
filtering fordiscrete-time singular Markovian jump systems with time-varying delay and partially unknown transition probabilitiesrdquoSignal Processing vol 91 no 2 pp 277ndash289 2011
[39] J Tian Y Li J Zhao and S Zhong ldquoDelay-dependent stochasticstability criteria for Markovian jumping neural networks withmode-dependent time-varying delays and partially knowntransition ratesrdquo Applied Mathematics and Computation vol218 no 9 pp 5769ndash5781 2012
[40] Y Wei J Qiu H R Karimi and M Wang ldquoA new design 119867infin
filtering for contrinuous-time Markovian jump systems withtime-varying delay and partially accessible mode informationrdquoSignal Processing vol 93 pp 2392ndash2407 2013
[41] L Zhang E-K Boukas and J Lam ldquoAnalysis and synthesisof Markov jump linear systems with time-varying delays andpartially known transition probabilitiesrdquo IEEE Transactions onAutomatic Control vol 53 no 10 pp 2458ndash2464 2008
[42] L Zhang and E-K Boukas ldquoMode-dependent 119867infin
filteringfor discrete-time Markovian jump linear systems with partlyunknown transition probabilitiesrdquo Automatica vol 45 no 6pp 1462ndash1467 2009
[43] L Zhang and E-K Boukas ldquoMode-dependent 119867infin
controlfor discrete-time Markovian jump linear systems with partlyunknown transition probabilitiesrdquo International Journal ofRobust and Nonlinear Control vol 19 no 8 pp 868ndash883 2009
[44] Y Zhang Y He M Wu and J Zhang ldquoStabilization forMarkovian jump systems with partial information on transi-tion probability based on free-connection weighting matricesrdquoAutomatica vol 47 no 1 pp 79ndash84 2011
[45] G Zong D Yang L Hou and Q Wang ldquoRobust finite-time119867infincontrol for Markovian jump systems with partially known
transition probabilitiesrdquo Journal of the Franklin Institute vol350 no 6 pp 1562ndash1578 2013
[46] Y Ding H Zhu S Zhong and Y Zhang ldquo1198712minus 119871infin
filteringfor Markovian jump systems with time-varying delays andpartly unknown transition probabilitiesrdquo Communications inNonlinear Science and Numerical Simulation vol 17 no 7 pp3070ndash3081 2012
[47] Q Ma S Xu and Y Zou ldquoStability and synchronizationfor Markovian jump neural networks with partly unknowntransition probabilitiesrdquo Neurocomputing vol 74 pp 3403ndash3411 2011
[48] E Tian D Yue and G Wei ldquoRobust control for Markovianjump systems with partially known transition probabilities andnonlinearitiesrdquo Journal of the Franklin Institute vol 350 no 8pp 2069ndash2083 2013
[49] Y Guo and Z Wang ldquoStability of Markovian jump systemswith generally uncertain transition ratesrdquo Journal of the FranklinInstitute vol 350 no 9 pp 2826ndash2836 2013
[50] L Zhang ldquo119867infinestimation for discrete-time piecewise homoge-
neous Markov jump linear systemsrdquo Automatica vol 45 no 11pp 2570ndash2576 2009
[51] Z Wu H Su and J Chu ldquostate estimation for discreteMarkovian jumping neural networks with time delayrdquo Neuro-computing vol 73 pp 2247ndash2254 2010
[52] I R Petersen and A V Savkin Robust Kalman Filteringfor Signals and Systems with Large Uncertainties BirkhauserBoston Mass USA 1999
[53] B D O Anderson and J B Moore Optimal Filtering Prentice-Hall Upper Saddle River NJ USA 1979
[54] I R Petersen and D C McFarlane ldquoOptimal guaranteedcost control and filtering for uncertain linear systemsrdquo IEEETransactions on Automatic Control vol 39 no 9 pp 1971ndash19771994
[55] S Xu T Chen and J Lam ldquoRobust 119867infin filtering for uncertainMarkovian jump systems with mode-dependent time delaysrdquoIEEE Transactions on Automatic Control vol 48 no 5 pp 900ndash907 2003
[56] J H Kim D Oh and H Park ldquoGuaranteed cost and 119867infin
filtering for time-delay systemsrdquo in Proceeding of the AmericanControl Conference pp 4014ndash4018 IEEE Arlington Va USA2001
[57] L Xie ldquoOutput feedback119867infincontrol of systems with parameter
uncertaintyrdquo International Journal of Control vol 63 no 4 pp741ndash750 1996
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Abstract and Applied Analysis 11
[35] B Du J Lam Y Zou and Z Shu ldquoStability and stabilization forMarkovian jump time-delay systems with partially unknowntransition ratesrdquo IEEE Transactions on Circuits and Systems IRegular Papers vol 60 no 2 pp 341ndash351 2013
[36] L Zhang and J Lam ldquoNecessary and sufficient conditionsfor analysis and synthesis of Markov jump linear systemswith incomplete transition descriptionsrdquo IEEE Transactions onAutomatic Control vol 55 no 7 pp 1695ndash1701 2010
[37] Y Guo and F Zhu ldquoNew results on stability and stabilizationof Markovian jump systems with partly known transitionprobabilitiesrdquoMathematical Problems in Engineering vol 2012Article ID 869842 11 pages 2012
[38] J Lin S Fei and J Shen ldquoDelay-dependent 119867infin
filtering fordiscrete-time singular Markovian jump systems with time-varying delay and partially unknown transition probabilitiesrdquoSignal Processing vol 91 no 2 pp 277ndash289 2011
[39] J Tian Y Li J Zhao and S Zhong ldquoDelay-dependent stochasticstability criteria for Markovian jumping neural networks withmode-dependent time-varying delays and partially knowntransition ratesrdquo Applied Mathematics and Computation vol218 no 9 pp 5769ndash5781 2012
[40] Y Wei J Qiu H R Karimi and M Wang ldquoA new design 119867infin
filtering for contrinuous-time Markovian jump systems withtime-varying delay and partially accessible mode informationrdquoSignal Processing vol 93 pp 2392ndash2407 2013
[41] L Zhang E-K Boukas and J Lam ldquoAnalysis and synthesisof Markov jump linear systems with time-varying delays andpartially known transition probabilitiesrdquo IEEE Transactions onAutomatic Control vol 53 no 10 pp 2458ndash2464 2008
[42] L Zhang and E-K Boukas ldquoMode-dependent 119867infin
filteringfor discrete-time Markovian jump linear systems with partlyunknown transition probabilitiesrdquo Automatica vol 45 no 6pp 1462ndash1467 2009
[43] L Zhang and E-K Boukas ldquoMode-dependent 119867infin
controlfor discrete-time Markovian jump linear systems with partlyunknown transition probabilitiesrdquo International Journal ofRobust and Nonlinear Control vol 19 no 8 pp 868ndash883 2009
[44] Y Zhang Y He M Wu and J Zhang ldquoStabilization forMarkovian jump systems with partial information on transi-tion probability based on free-connection weighting matricesrdquoAutomatica vol 47 no 1 pp 79ndash84 2011
[45] G Zong D Yang L Hou and Q Wang ldquoRobust finite-time119867infincontrol for Markovian jump systems with partially known
transition probabilitiesrdquo Journal of the Franklin Institute vol350 no 6 pp 1562ndash1578 2013
[46] Y Ding H Zhu S Zhong and Y Zhang ldquo1198712minus 119871infin
filteringfor Markovian jump systems with time-varying delays andpartly unknown transition probabilitiesrdquo Communications inNonlinear Science and Numerical Simulation vol 17 no 7 pp3070ndash3081 2012
[47] Q Ma S Xu and Y Zou ldquoStability and synchronizationfor Markovian jump neural networks with partly unknowntransition probabilitiesrdquo Neurocomputing vol 74 pp 3403ndash3411 2011
[48] E Tian D Yue and G Wei ldquoRobust control for Markovianjump systems with partially known transition probabilities andnonlinearitiesrdquo Journal of the Franklin Institute vol 350 no 8pp 2069ndash2083 2013
[49] Y Guo and Z Wang ldquoStability of Markovian jump systemswith generally uncertain transition ratesrdquo Journal of the FranklinInstitute vol 350 no 9 pp 2826ndash2836 2013
[50] L Zhang ldquo119867infinestimation for discrete-time piecewise homoge-
neous Markov jump linear systemsrdquo Automatica vol 45 no 11pp 2570ndash2576 2009
[51] Z Wu H Su and J Chu ldquostate estimation for discreteMarkovian jumping neural networks with time delayrdquo Neuro-computing vol 73 pp 2247ndash2254 2010
[52] I R Petersen and A V Savkin Robust Kalman Filteringfor Signals and Systems with Large Uncertainties BirkhauserBoston Mass USA 1999
[53] B D O Anderson and J B Moore Optimal Filtering Prentice-Hall Upper Saddle River NJ USA 1979
[54] I R Petersen and D C McFarlane ldquoOptimal guaranteedcost control and filtering for uncertain linear systemsrdquo IEEETransactions on Automatic Control vol 39 no 9 pp 1971ndash19771994
[55] S Xu T Chen and J Lam ldquoRobust 119867infin filtering for uncertainMarkovian jump systems with mode-dependent time delaysrdquoIEEE Transactions on Automatic Control vol 48 no 5 pp 900ndash907 2003
[56] J H Kim D Oh and H Park ldquoGuaranteed cost and 119867infin
filtering for time-delay systemsrdquo in Proceeding of the AmericanControl Conference pp 4014ndash4018 IEEE Arlington Va USA2001
[57] L Xie ldquoOutput feedback119867infincontrol of systems with parameter
uncertaintyrdquo International Journal of Control vol 63 no 4 pp741ndash750 1996
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of