Embed Size (px)
Citation preview
Accep
ted M
anus
cript
A hybrid direction algorithm for solving optimal control problems
Mohamed Abdelaziz Zaitri, Mohand Ouamer Bibi and Mohand Bentobache
Accepted Manuscript Version
This is the unedited version of the article as it appeared upon acceptance by the journal. A final edited version of the article in the journal format will be made available soon.
As a service to authors and researchers we publish this version of the accepted manuscript (AM) as soon as possible after acceptance. Copyediting, typesetting, and review of the resulting proof will be undertaken on this manuscript before final publication of the Version of Record (VoR). Please note that during production and pre-press, errors may be discovered which could affect the content.
© 2019 The Author(s). This open access article is distributed under a Creative Commons Attribution (CC-BY) 4.0 license.
Publisher: Cogent OA
Journal: Cogent Mathematics & Statistics
DOI: http://dx.doi.org/10.1080/25742558.2019.1612614
Accep
ted M
anus
cript
A hybrid direction algorithm for solving optimal control problems Mohamed Abdelaziz Zaitri1, Mohand Ouamer Bibi1 and Mohand Bentobache2 1Research Unit LaMOS (Modeling and Optimization of Systems),
University of Bejaia, 06000, Algeria 2LMPA, Laboratory of Pure and Applied Mathematics,
University of Laghouat, 03000, Algeria [email protected], [email protected],
21st February 2019
Abstract
In this paper, we present an algorithm for finding an approximate numerical solution for linear optimal control problems. This algorithm is based on the hybrid direction algorithm developed by Bibi and Bentobache [A hybrid direction algorithm for solving linear programs, International Journal of Computer Mathematics, vol. 92, no.1, pp. 201-216, 2015]. We define an optimality estimate and give a necessary and sufficient condition to characterize the optimality of a certain admissible control of the discretized problem, then we give a numerical example to illustrate the proposed approach. Finally, we present some numerical results which show the convergence of the proposed algorithm to the optimal solution of the presented continuous optimal control problem. Keywords: Optimal control, linear programming, hybrid direction algorithm, optimality estimate, numerical results.
1 Introduction The optimal control theory consists in finding a control which optimizes a functional on a domain described by a system of differential equations, with box and terminal constraints on the control. This theory is applied in various fields of the engineering sciences: aeronautics, physics, finance, etc. Because of the importance of this theory, several researchers have been interested in the development of effective numerical methods for solving this type of problems. In [9, 10], the authors developed the adaptive method for solving linear optimal control problems. This method is then generalized for solving general quadratic optimization problems [4, 5, 8, 11, 12].
In [2, 3, 6, 7], the authors proposed a new improvement direction for the adaptive method in order to solve linear programming problems with bounded variables. This direction is called hybrid direction because some of its components take extreme values and the other components take the values of the opposite gradient.
In this paper, we present an algorithm based on this hybrid direction for solving linear optimal control problems. In a similar way to [7], we define an optimality estimate and give a necessary and sufficient condition to characterize the optimality of a certain admissible control of the discretized problem. Then we describe a numerical algorithm for finding an approximate solution and we present some numerical results in order to show its convergence.
The paper is organized as follows : In Section 1, we present the problem and give some definitions. In Section 2, we present the details of the proposed algorithm and we give a
Accep
ted M
anus
cript
numerical example to illustrate our approach. Finally, we conclude this paper and give some perspectives.
2 Optimal control problem
2.1 State of the problem and definitions
Consider the following terminal optimal control problem:
*
* 0
*
* ** *
( ) = ( ) , (1.1)max
= ( ) = ( ) ( ), ( ) = , (1.2)
( ) = , (1.3)
( ) , = [ , ], (1.4)
uJ u c x t
dxx t Ax t bu t x t x
dtHx t g
f u t f t T t t
′ → + ≤ ≤ ∈
(1)
where ( )J u is the quality criterion, ( )n nA ×∈ is the dynamic matrix of the system,
( ) nx t ∈ is the state vector of the system, nb ∈ , ( )m nH ×∈ is a matrix of rank m n≤ , mg ∈ , ( )u t ∈ is a piecewise constant control bounded by *
*,f f ∈ and nc ∈ . The
symbol (’) designates the transposition operation. Definition 1 Any control = ( ( ), )u u t t T∈ satisfying the constraints:
* * ** *( ) , = [ , ], ( ) = ,f u t f t T t t Hx t g≤ ≤ ∈
is called an admissible control of the problem (1) .
An admissible control 0 0= ( ( ), )u u t t T∈ is said to be optimal if
0( ) = ( ).maxu
J u J u
An admissible control u ε is said to be ε − optimal or suboptimal if 0( ) ( ) ,J u J uε ε− ≤
where 0ε ≥ is an accuracy chosen in advance. The solution of the problem consists in the determination of an admissible control 0u
which, with the trajectory 0x , maximizes the quality criterion ( )J u :
0 * 0 *( ) = ( ) = ( ) = ( ).max maxu u
J u J u c x t c x t′ ′
The solution of the system (1.2) is given by
10
*
( ) = ( ) ( ) ( ) , ,t
t
x t F t x F bu d t Tτ τ τ− + ∈
(2)
where ( ), ,F t t T∈ is the solution of the system
*
( ) = ( ),
( ) = , ,n
F t AF t
F t I t T
∈
and the matrix nI represents the identity matrix of order n .
Accep
ted M
anus
cript
By replacing the expression (2) in the system (1.1) (1.3)− , we find
*
* * * 10
*
( ) = ( ) = ( ) ( ) ( ) ( ) ,t
t
J u c x t c F t x c F t F t bu t dt−′ ′ ′+
*
* * 10
*
( ) ( ) ( ) ( ) = .t
t
HF t x HF t F t bu t dt g−+
If we set * 1( ) = ( ) ( ) ,c t c F t F t b−′
* 1( ) = ( ) ( ) ,t HF t F t bφ −
*
0 0= ( ) ,g g HF t x−
then we get the following equivalent problem:
*
*0
*
*
0
*
* ** *
( ) = ( ) ( ) ( ) , (3.1)max
( ) ( ) = , (3.2)
( ) , = [ , ]. (3.3)
t
ut
t
t
J u c F t x c t u t dt
t u t dt g
f u t f t T t t
φ
′ + →
≤ ≤ ∈
(3)
2.2 Discretization of the initial problem
We choose a subset ** *= { , , , },hT t t h t h+ − where
**=
t th
N
− and *.N ∈ Let the function
( ),u t t T∈ , be a piecewise constant control such that
( ) ( ), [ , [, .hu t u t h Tτ τ τ τ≡ ∈ + ∈
Using this discretization, the problem (3.1) (3.3)− becomes:
*0
0
**
( ) = ( ) ( ) ( ) , (4.1)max
( ) ( ) = , (4.2)
( ) , , (4.3)
uTh
Th
h
J u c F t x q u
d u g
f u f T
τ
τ
τ τ
τ τ
τ τ
∈
∈
′ + → ≤ ≤ ∈
(4)
where
( ) = ( ) and ( ) = ( ) .h h
q c s ds s dsτ τ
τ τ
τ τ φ+ +
(5)
2.3 Support control
The set = { , = 1, } ,B i hT i m Tτ ⊂ is called a support if the corresponding matrix
Accep
ted M
anus
cript
= ( ( ), ) ( )B B m mP d Tτ τ ×∈ ∈ is nonsingular.
The pair { , }Bu T formed by the admissible control u and the support BT is called a support
control of the problem (4) . The latter is said to be nondegenerate if ** < ( ) < , .Bf u f Tτ τ ∈
2.4 Increment formula of the functional Let { , }Bu T be a support control and ( ), ,x t t T∈ its corresponding trajectory. Using the support
BT , we construct the vector of the potentials mν ∈ and the cocontrol vector ( ), hE Tτ τ ∈ , as
follows: 1= , ( ) = ( ) ( ),B Bq P E d qν τ ν τ τ−′ ′ ′ − (6)
where = ( ( ), )B Bq q Tτ τ ∈ , = ( ( ), ) = 0B BE E Tτ τ ∈ .
Consider another control
( ) = ( ) ( ), ,u t u t u t t T+ Δ ∈ and the corresponding trajectory
( ) = ( ) ( ), .x t x t x t t T+ Δ ∈ Then the increment of the functional (4.1) is given by
( ) = ( ) ( ) = ( ) ( ), = \ .N h BTN
J u J u J u E u T T Tτ
τ τ∈
Δ − − Δ
The following theorem gives a necessary and sufficient condition of optimality for an
admissible control u of the problem (4) . Theorem 1 [10] The following relationships:
**
**
( ) = , ( ) > 0,
( ) = , ( ) < 0,
( ) , ( ) = 0, ,N
u f if E
u f if E
f u f if E T
τ ττ τ
τ τ τ
≤ ≤ ∈
are sufficient, and in the case of the nondegeneracy of the support control { , }Bu T also
necessary, for the optimality of the admissible control .u
3 An iteration of the hybrid direction algorithm Let { , }Bu T be a support control for the problem (4) and [0,1]η ∈ . Define the following sets:
( )*= { : ( ) > ( ) },N NT T E u fτ τ η τ+ ∈ −
( )*= { : ( ) < ( ) },N NT T E u fτ τ η τ− ∈ −
( )*= { : 0 < ( ) ( ) },PN NT T E u fτ τ η τ
+∈ ≤ −
( )*= { : ( ) ( ) < 0},PN NT T u f Eτ η τ τ
−∈ − ≤
0
= { : ( ) = 0},PN NT T Eτ τ∈
Accep
ted M
anus
cript
( ) ( ) 0**= { : ( ) ( ) ( ) } = .P P P P
N N N N NT T u f E u f T T Tτ η τ τ η τ+ −
∈ − ≤ ≤ − ∪ ∪
Then = .P
h N N N BT T T T T+ −∪ ∪ ∪
Recall that the suboptimality estimate ( , )Bu Tβ is given by the following formula [10]:
**( , ) = ( )( ( ) ) ( )( ( ) ),B
T T
u T E u f E u fτ τ
β τ τ τ τ+ −∈ ∈
− + −
where = PN NT T T
++ + ∪ and = .PN NT T T
−− − ∪
We call optimality estimate, the quantity ( , , )Bu Tγ η defined by:
* 2*
1( )( ( ) ) ( )( ( ) ) ( ), if > 0,
( , ), if = 0.
T T P PT TN N N N
B
E u f E u f E
u T
τ τ τ
τ τ τ τ τ ηη
β η
+ − + −∈ ∈ ∈ ∪
− + − +
(7)
Theorem 2 (Necessary and sufficient condition of optimality [7]) Let { , }Bu T be a support control for the problem (4) and > 0η . Then the condition
( , , ) = 0Bu Tγ η is sufficient and, in the case of the nondegeneracy of the support control { , }Bu T
also necessary, for the optimality of the admissible control u . Let { , }Bu T be a starting support control of the problem (4) , for which the optimality
criterion is not satisfied. An iteration of the hybrid direction algorithm consists in moving from
{ , }Bu T to { , }Bu T , where 0= .u u uθ+ Δ This passage is done in two steps:
1. Change of control: .u u→
2. Change of support: .BBT T→
3.1 Change of control Let { , }Bu T be a support control for the problem (4) and [0,1]η ∈ . We compute ( , , )Bu Tγ η with
the formula (7). If ( , , ) = 0,Bu Tγ η then the support control { , }Bu T is optimal, otherwise we
define the admissible improvement direction ( )u τΔ as follows:
**
( ), for ,
( ), for ,
( )( ) = , for , 0,
0, for , = 0,
, for ,
N
N
PN
PN
B B
f u T
f u T
Eu T
T
u T
τ ττ τ
ττ τ ηη
τ ητ
+
−
− ∈ − ∈−Δ ∈ ≠ ∈
Δ ∈
(8)
where 1= ,B B N Nu P P u−Δ − Δ = ( ( ), )N NP d Tτ τ ∈ and = ( ( ), ).N Nu u Tτ τΔ Δ ∈ This direction is
called a hybrid direction [7]. The direction ( )u τΔ is an admissible one for the problem (4) . Indeed,
Accep
ted M
anus
cript
0( ) ( ) = ( ) ( ) ( ) ( )T T Th h h
d u d u d uτ τ τ
τ τ τ τ θ τ τ∈ ∈ ∈
+ Δ
0= ( ) ( ) ( )N N B BTh
d u P u P uτ
τ τ θ∈
+ Δ + Δ
0 1= ( ) ( ) ( )N N B B N NTh
d u P u P P P uτ
τ τ θ −
∈
+ Δ − Δ
= ( ) ( )Th
d uτ
τ τ∈
0= .g
To improve the objective function while remaining within the admissible domain, we
compute the step 0θ along the direction ( )u τΔ :
01= min{ ( ),1},θ θ τ (9)
where 1( ) = min{ ( ), },BTθ τ θ τ τ ∈ with
*
*
( ), ( ) > 0,
( )
( )( ) = , ( ) < 0,
( )
, ( ) = 0.
f uif u
u
f uif u
u
if u
τ τττθ τ τ
ττ
− Δ Δ − Δ Δ∞ Δ
(10)
Then the new admissible control will be: 0( ) = ( ) ( ),u u uτ τ θ τ+ Δ (11)
where ( )u τΔ and 0θ are defined by relationships (8) and (9) respectively. The increment of the objective function is then 0( ) = ( ) ( )
t TN
J u E uθ τ τ∈
Δ − Δ
0 0 0= ( ) ( ) ( ) ( ) ( ) ( )T T p PN N T TNN
E t u E u E uτ τ τ
θ τ θ τ τ θ τ τ+ − + −∈ ∈ ∈ ∪
− Δ − Δ − Δ
2
0 0 * 0*
( )= ( )( ( ) ) ( )( ( ) )
T T p pN N T TN N
EE u f E u f
τ τ τ
τθ τ τ θ τ τ θη+ − + −∈ ∈ ∈ ∪
− + − +
0= ( , , ) 0.Bu Tθ γ η ≥
So ( ) > ( )J u J u , for 0 > 0θ . Corollary 1 [7]
If 0 = 1θ and = ,P PN NT T
+ −∪ ∅ then u is optimal.
3.2 Change of support
Accep
ted M
anus
cript
If for the support control { , }Bu T of the problem (4) , we have 0 < 1,θ then we change BT by BT
using the dual method. For this, we compute the vector ω and the number 0α as follows:
0 1 1( ) = ( ) ( ), and = ( ) ( ).hu u T uω τ τ τ τ α ω τ τ+ Δ ∈ −
Then the new cocontrol will be given by: 0( ) = ( ) ( ), ,hE E Tτ τ σ δ τ τ+ ∈
where ( )δ τ is the dual direction and 0σ the dual step, which are computed as follows:
1
0 1
0 11
0, if \ { },
1, if > 0, = ,( ) =
1, if < 0, = ,
( ), for , = ( ( ), ),
B
B B N B B
T
P d T T
τ τα τ τ
δ τα τ τ
δ τ τ δ δ τ τ−
∈−+ ′ ′∈ ∈
(12)
0
0= ( ) = ( ),minTNτ
σ σ τ σ τ∈
(13)
where
*
*
( ), if ( ) ( ) < 0,
( )
0, if ( ) = 0, ( ) < 0, ( ) ,( ) =
0, if ( ) = 0, ( ) > 0, ( ) ,
, elsewhere, .N
EE
E f
E f
T
τ τ δ τδ τ
τ δ τ ω τσ ττ δ τ ω τ
τ
− ≠ ≠
+∞ ∈
The following new support is then obtained: 1 0= ( \ { }) { }.B BT T τ τ∪
3.3 Scheme of the hybrid direction algorithm Let { , }Bu T be a support control for the problem (4) and η a real number such that [0,1]η ∈ . In
order to take into account the specificity of the studied linear optimal control problem, we present in this section a slightly modified version of the algorithm presented in [7]. Indeed, if
0 = 1θ and P PN NT T
+ −∪ ≠ ∅ , then we reduce the value of the parameter η by setting = / 2η η
and we start a new iteration with the new control u . The scheme of the hybrid direction algorithm for solving the linear optimal control problem is described in the following steps: Algorithm 1 (1) Compute ( ), ( ), , ( )d q Eτ τ ν τ with relationships (5)-(6);
(2) Determine the sets NT + , NT − , PNT
+ and P
NT−
;
(3) Compute ( , , )Bu Tγ η with the formula (7);
(4) If ( , , ) = 0Bu Tγ η , then the algorithm stops with { , }Bu T , an optimal support control for the
discretized problem; (5) Compute the improvement direction ( )u τΔ using the relationship (8);
Accep
ted M
anus
cript
(6) Compute 1( ) = ( ),minTBτ
θ τ θ τ∈
where ( )θ τ is determined by (10);
(7) Compute 01= min{1, ( )},θ θ τ 0( ) = ( ) ( ), ,hu u u Tτ τ θ τ τ+ Δ ∈ and
0( ) = ( ) ( , , )BJ u J u u Tθ γ η+ ;
(8) If 0 = 1θ , then
(8.1) - If =P PN NT T
+ −∪ ∅ , then ( )u τ is optimal. Stop.
(8.2) - Else, set = / 2η η , ( ) = ( )u uτ τ , ( ) = ( )J u J u and go to step (2).
(9) Compute the dual direction ( ), ,hTδ τ τ ∈ using the relationship (12);
(10) Compute the dual step 0σ and determine 0τ using the relationship (13);
(11) Set 0( ) = ( ) ( ), ,hE E Tτ τ σ δ τ τ+ ∈ 1 0= ( \ { }) { }B BT T τ τ∪ ;
(12) Set ( ) = ( )u uτ τ , =B BT T , ( ) = ( )J u J u , ( ) = ( )E Eτ τ and go to step (2).
4 Numerical example Consider the following problem
( ) = (2) ,max
( ) = ( ) ( ), (0) = 0,
(2) = ,
1 ( ) 1, = [0,2],
uJ u c x
x t Ax t bu t x
Hx g
u t t T
′ → +− ≤ ≤ ∈
(14)
with
0 1 0 01
= 0 0 , = 1 , = (1, 2), = , = 1 .2
A b H g c
−
We have
1
1 1
( ) = 0 1 , ( ) = 0 1 ,
t t
F t F t−
−
1 1( ) = (2) ( ) = 1, ( ) = (2) ( ) = .c t c F F t b t HF F t b tφ− −′ −
Consider the admissible control
1, if [0,1[
2( ) =1
, if [1,2].2
tu t
t
∈− ∈
The corresponding trajectory is
Accep
ted M
anus
cript
2
4
( ) = , if [0,1[2
t
tx t t
∈
and
2 1
4 2
( ) = 1 , if [1,2].2
tt
tx t t
− + −
− + ∈
We choose = 0.5h and = 1.η We take the support control { , }Bu T for the problem (14), with
= {1}BT .
Iteration 1.
3 3
2 21
1 1
5 8 1= = , = , = = ,
8 5 2B B BP sds P q ds−− − −
1 8 4 2
= = , ( ) = ( 1),2 5 5 5
Eν τ τ−′ × − −
1 3
= {0, , },2 2NT
( )2= { : ( 1) > ( ) 1 } = ,
5N NT T uτ τ τ+ ∈ − + ∅
( )2 3= { : 0 < ( 1) ( ) 1 } = { },
5 2P
N NT T uτ τ τ+
∈ − ≤ +
( )2= { : ( 1) < ( ) 1 } = ,
5N NT T uτ τ τ− ∈ − − ∅
( ) 2 1= { : ( ) 1 ( 1) < 0} = {0, }.
5 2P
N NT T uτ τ τ−
∈ − ≤ −
We compute the optimality estimate:
2 2 21 3 6( , , ) = (0) ( ) ( ) = ,
2 2 25Bu T E E Eγ η + +
so the control u is not optimal. Change of control: We compute ( )u τΔ :
Accep
ted M
anus
cript
2, if = 0,
51 1
, if = ,5 2( ) =2
, if = 1,251 3
, if = .5 2
u
τ
ττ
τ
τ
++
Δ +−
The step along the direction ( )u τΔ is computed as follows:
We have 2
(1) = ,25
uΔ then 1
11 752( ) = =
2 425
θ τ−
and 0 = 1.θ
Hence, the new control is given by:
9, if = 0,
107 1
, if = ,10 2( ) =21
, = 1,507 3
, if = .10 2
u
if
τ
ττ
τ
τ
++−−
We have 0 = 1θ and ,P PN NT T
+ −∪ ≠ ∅ so we set
2
= / 2 = 1 / 2, ( ) = ( 1), ( ) = ( ).5
E u uη η τ τ τ τ−
Iteration 2. For this iteration, we have
3 1 4
= { }, = {0, }, = and ( , , ) = .2 2 25
PN N N BT T u Tγ η+ − ∅
We compute the direction ( )u τΔ :
1, if = 0,
103 1
, if = ,10 2( ) =11
, = 1,503 3
, if = .10 2
u
if
τ
ττ
τ
τ
++
Δ +−
Accep
ted M
anus
cript
We have 11
(1) = ,50
uΔ hence 1
211 7150( ) = =
11 1150
θ τ+
and 0 = 1.θ So
1, if = 0,
11, if = ,
2( ) = 1
, if = 1,5
31, if = .
2
u
τ
τ
ττ
τ
++−−
Since 0 = 1θ and = ,P PN NT T
+ −∪ ∅ then the control
0
1, if = 0,
11, if = ,
2( ) = 1
, if = 1,5
31, if = ,
2
u
τ
τ
ττ
τ
++−−
is optimal for the discretized problem, with 0 2( ) =
5J u . Therefore, the control
1, if [0,1[
1 3( ) = , if [1, [
5 23
1, if [ ,2],2
t
u t t
t
∈− ∈− ∈
is an approximate solution of the original problem (14). In order to find a good approximate solution for the original continuous problem (14),
we have implemented the discretization technique using the Cauchy formula and the hybrid direction algorithm with MTALB2018a. The developed solver was tested on a computer surface pro 2, with 4GO of memory and processor Intel(R) Core(TM) i5-4300U CPU 1.90GHz 2.50GHz, running under Microsoft Window 10 operating system.
The initialization approach proposed in [1] can be used to compute an initial admissible support control, however we have initialized the hybrid direction algorithm with the following obvious admissible control:
Accep
ted M
anus
cript
1, if [0,1[
21
( ) = , if [1,2].2
t
u t t
∈− ∈
In Table 1, we report numerical results for different values of N , where 1CPU , 2CPU ,
IT and 0J represent respectively the cpu time of the discretization phase, the execution time, the number of iterations of the hybrid direction algorithm and the optimal value of the quality criterion of (14). We plot the optimal control in terms of t for = 5000N and we plot the optimal objective values of the linear program (4) corresponding to the problem (14) in terms of N . Table 1: Numerical simulation results for the problem (14).
N 1CPU 2CPU IT 0J
10 0.2689 0.0397 9 0.4461538462 50 1.0252 0.0177 38 0.4491803279 100 1.5805 0.0191 71 0.4494308943 200 3.3212 0.0175 134 0.4494693878 500 7.9160 0.0492 320 0.4494877651 1000 16.1820 0.1482 628 0.4494889796 2000 32.6202 0.5174 1243 0.4494895876 3000 47.3245 1.0210 1856 0.4494897052 5000 80.9802 3.2498 3082 0.4494897273
Accep
ted M
anus
cript
Figure 1: Graph of the optimal control in terms of t for = 5000N .
Figure 2: Graph of the objective function value in terms of N .
Accep
ted M
anus
cript
Note that for large values of N , our method converges to the optimal value of the continuous original problem * = 0.4495J . Furthermore, we can see from Graph of Figure 1, that the commutation time is approximately equal to 1.23 sec.
5 Conclusion In this paper, we applied the hybrid direction algorithm developed in [7] to find an approximate optimal solution to a linear optimal control problem. A numerical example was given to illustrate the described algorithm and some numerical simulation results were presented in order to show the convergence of our algorithm to the optimal solution of the continuous problem. In a future work, we will compare the presented approach with classical approaches on practical optimal control problems.
References [1] M. Bentobache and M. O. Bibi, A Two-phase Support Method for Solving Linear
Programs: Numerical Experiments, Mathematical Problems in Engineering, vol. 2012, Article ID 482193, 28 pages doi:10.1155/2012/482193.
[2] M. Bentobache, On mathematical methods of linear and quadratic programming,
PhD thesis, University of Bejaia, 2013. [3] M. Bentobache and M.O. Bibi, Numerical methods of linear and quadratic
programming, French Academic Editions, Germany, 2016 (in French) [4] M. O. Bibi, Optimization of a linear dynamic system with double terminal constraints
on the trajectories, Optimization, vol. 30, no. 4, pp. 359–366, 1994. [5] M. O. Bibi, Support method for solving a linear-quadratic problem with polyhedral
constraints on control, Optimization, vol. 37, no. 2, pp. 139–147, 1996. [6] M. O. Bibi and M. Bentobache, The adaptive method with hybrid direction for
solving linear programming problems with bounded variables, In: Proceedings of COSI’2011, University of Guelma, Algeria, pp. 80-91, 24-27 April 2011.
[7] M. O. Bibi and M. Bentobache, A hybrid direction algorithm for solving linear
programs, International Journal of Computer Mathematics, vol. 92, no. 1, pp. 201-216, 2015. [8] B. Brahmi and M. O. Bibi, Dual support method for solving convex quadratic
programs, Optimization, Vol. 59, No. 6, 2010, pp. 851–872. [9] R. Gabasov and F. M. Kirillova, Methods of linear programming, Vol. 1, 2 and 3,
Edition of the Minsk University, 1977, 1978 and 1980 (in Russian). [10] R. Gabasov, F.M. Kirillova and S.V. Prischepova, Optimal Feedback Control,
Springer-Verlag, London, 1995.
Accep
ted M
anus
cript
[11] N. Khimoum and M. O. Bibi, Primal-dual method for solving a linear-quadratic
multi-input optimal control problem, Optimization letters, https://doi.org/10.1007/s11590-018-1375-2, 2019.
[12] E. A. Kostina and O. I. Kostyukova, An algorithm for solving quadratic
programming problems with linear equality and inequality constraints, Computational Mathematics and Mathematical Physics, vol. 41, no. 7, pp. 960-973, 2001. Biography of the authors:
Mohamed Abdelaziz Zaitri received a DES degree in Mathematics from the Higher Normal School of Kouba (2011), Algeria, a master degree (2014) from the University of Bejaia. He is a PhD student at the Department of Operational Research, University of Bejaia. Mohand Ouamer Bibi received a DES degree in Mathematical Analysis (1980) from the USTHB University, Algeria, a master degree (1982) and PhD (1986) in Applied Mathematics from the University of Minsk, Belarus. He is a Full Professor at the Department of Operational Research and the leader of the research group “Optimization and Optimal Control” at the LaMOS Research Unit, University of Bejaia, Algeria. Mohand Bentobache received his engineering degree in operational research (2002), master degree (2005), PhD (2013) and HDR (2015) in Applied Mathematics from the University of Bejaia, Algeria. He is an Associate Professor at the Department of Technology and the leader of the research group “Numerical Analysis and Optimization” at the Laboratory of LMPA, University of Laghouat, Algeria. Description of the work for the large public readers:
The optimal control theory consists in finding a control which optimizes a functional on a domain described by a system of differential equations, with box and terminal constraints on the control. This theory is applied in various fields of the engineering sciences: aeronautics, physics, finance, etc. For example, finding the minimal time necessary for moving a missile from one starting point to a destination point can be modeled as a linear optimal control, where the constraints are given by the motion equations of the missile. In this work, we have proposed a method which finds a numerical solution for the linear optimal problem. Our method can be used for the simulation of optimal trajectories of control problems which arises in military applications, finance, etc.
