Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
LINEAR CONTROLLINEAR CONTROLSYSTEMSSYSTEMS
Ali KarimpourAssociate Professor
Ferdowsi University of Mashhad
Dr. Ali Karimpour Nov 2014
Lecture 4
2
Lecture 4
Topics to be covered include:
External Description.(TF Model)
Internal Model Description. (SS Model)
External and Internal Description Model. (SS and TF Description)
Dr. Ali Karimpour Nov 2014
Lecture 4
3
Topics to be covered include:
External Description.(TF Model)
Internal Model Description. (SS Model)
External and Internal Description Model. (SS and TF Description)
Dr. Ali Karimpour Nov 2014
Lecture 4
4
Property.
Poles and zeros and their physical meaning.
SISO and MIMO.
Open loop and closed loop system.
Effect of feedback .
Characteristic equation for SISO system
External Description (Transfer function representation)
Dr. Ali Karimpour Nov 2014
Lecture 4
5
TF model properties
خواص مدل تابع انتقال1- It is available just for linear time invariant systems. (LTI)
2- It is derived by zero initial condition.
3- It just shows the relation between input and output so it may lose some information.
4- It is just dependent on the structure of system and it is independent to the input value and type.
5- It can be used to show delay systems but SS can not.
Dr. Ali Karimpour Nov 2014
Lecture 4
6
Table : Laplace transform table
جدول تبدیل الپالس
u(t) - u(t-τ)
Dr. Ali Karimpour Nov 2014
Lecture 4
7
Table : Laplace transform properties.
خواص تبدیل الپالس
y(t-τ) u(t-τ)
Dr. Ali Karimpour Nov 2014
Lecture 4
8
Laplace transform properties. خواص تبدیل الپالس
Example 1: Check F.V.T and I.V.T for following:
)3(3)()
ss
sYa Both F.V.T and I.V.T are valid.
2
1)()s
sYb Only I.V.T is valid.
12)()2
sssYc Only I.V.T is valid.
11)()
s
sYd Only I.V.T is valid.
1)()
sssYe I.V.T is valid with care.
Dr. Ali Karimpour Nov 2014
Lecture 4
9
Poles of Transfer functionقطب تابع انتقال
Pole: s=p is a value of s that the absolute value of TF is infinite.
How to find it?
)()()(sdsnsG Transfer function
Let d(s)=0 and find the roots
Why is it important? It shows the behavior of the system.Why?????
Dr. Ali Karimpour Nov 2014
Lecture 4
10
Example 2: Poles and their physical meaning.
656)( 2
ssssG
31
221
)65(6)(ofResponseStep 2
ssssssssc
S2+5s+6=0 p1= -2, p2=-3
Transfer functionroots([1 5 6])
)(1)(2)(1)( 32 tuetuetutc tt
p1= -2, p2=-3
قطبها و خواص فیریکی آنها
Dr. Ali Karimpour Nov 2014
Lecture 4
11
656)( 2
ssssGTransfer function
step([1 6],[1 5 6])
0 0.5 1 1.5 2 2.5 30
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1Step Response
Time (sec)
Ampl
itude
Example 2: Poles and their physical meaning(Continue).
Dr. Ali Karimpour Nov 2014
Lecture 4
12
Example 3: Derive the poles of Y(s).
)()2sin(1)( 22 tutteteeety tttt
-2 -1 1
2
-2
Remark1: Which terms go to infinity?
Remark 2: Poles on the imaginary axis?
Dr. Ali Karimpour Nov 2014
Lecture 4
13
Zeros of Transfer function صفر تابع انتقال
Zero: s=z is a value of s that the absolute value of TF is zero.
Who to find it?
)()()(sdsnsG Transfer function
Let n(s)=0 and find the roots
Why is it important? It also shows the behavior of system.Why?????
Dr. Ali Karimpour Nov 2014
Lecture 4
14
Example 4: Zeros and their physical meaning.
65610)(
656)( 2221
ssssG
ssssG
)(121)(3
12
21)65(
6)( 32121 tueetc
ssssssssc tt
Transfer functionsz1= -6 z2=-0.6
)(871)(3
82
71)65(
610)( 32222 tueetc
ssssssssc tt
صفرها و خواص فیریکی آنها
Remark: There is one more zero at infinity.
21
21
21 632
656)(
zzpp
ssssG Number of poles and zeros?
Dr. Ali Karimpour Nov 2014
Lecture 4
15
Transfer function65
610)(65
6)( 2221
ssssG
ssssG
step([1 6],[1 5 6]) ;hold on;step([10 6],[1 5 6])
صفرها و خواص فیریکی آنها
Example 4: Zeros and their physical meaning(Continue).
Dr. Ali Karimpour Nov 2014
Lecture 4
16
uuyyy 65
Transfer function of the system is:
651
)()(
2
ss
ssusy
It has a zero at z=-1
What does it mean? u=e-1t and suitable initial condition leads to y(t)=0
صفرها و خواص فیریکی آنها
Example 5: Zeros and their physical meaning.
Dr. Ali Karimpour Nov 2014
Lecture 4
1717
Open loop and closed loop systems
Open-loopControl system Controller
سیستمهاي حلقه باز و حلقه بسته
Actuator
Dr. Ali Karimpour Nov 2014
Lecture 4
1818
When is the open loop representation applicable?
The model on which the design of the controller has been based is a very good representation of the plant,
The model must be stable, and
Disturbances and initial conditions are negligible.
در چه شرایطی کنترل حلقه باز مناسب است؟
Dr. Ali Karimpour Nov 2014
Lecture 4
1919
Open loop and closed loop systems
Controller
Open-loopControl system Controller
Closed-loopControl system
Sensor
ActuatorActuator
Dr. Ali Karimpour Nov 2014
Lecture 4
2020
SISO and MIMO
سیستم تک ورودي تک خروجی و چند ورودي چند خروجی
Systemu1 y1
SISO
u1
u2
u3
y1
y2
y3
MIMO
System
Dr. Ali Karimpour Nov 2014
Lecture 4
2121
Sensitivity حساسیت
Sensitivity is defined as:
MG
GM
GGMMS
//
M is the system + controller transfer function
G is the system transfer function
Dr. Ali Karimpour Nov 2014
Lecture 4
2222
Sensitivity
)()( sKsGM
1)()(
)()(
sGsK
sGsKMG
GMS
)()(11
sGsKMG
GMS
حساسیت
G(s)K(s)
Open loop system
r c
G(s)K(s)+
-
r c
Closed loop system
)()(1)()()(sGsKsGsKsM
Dr. Ali Karimpour Nov 2014
Lecture 4
2323
Feedback properties
ویژگیهاي فیدبکFeedback can effect:
a) System gain
b) System stability
c) System sensitivity
d) Noise and disturbance
Dr. Ali Karimpour Nov 2014
Lecture 4
t (Time)
24
Example 6: A thermal system
یک سیستم حرارتی: 11مثال
Input signal is an step function so:
0001
)(tt
tus
su 1)(
000)(
ttkekty
t
/1)(
sk
sksy
Input signal1
Output signal
k
?)()()( susysg
1sk
Dr. Ali Karimpour Nov 2014
Lecture 4
25
Example 7: A system with pure time delay
سیستمی با تاخیر ثابت: 12مثال
1)(
sksg
dsTesh )(
1)()(
skesgsh
dsT
Dr. Ali Karimpour Nov 2014
Lecture 4
26
Topics to be covered include:
External Description.(TF Model)
Internal Model Description. (SS Model)
External and Internal Description Model. (SS and TF Description)
Dr. Ali Karimpour Nov 2014
Lecture 4
27
State space solution. State transition matrix. State transition equation.
Controllability and pole placement by state feedback. Observability by state feedback. Similarity transformation and canonical forms. Eigenvalues of the matrix A and poles of transfer function. Controllability and observability from block diagram.
Internal Description Model. (SS Description)
Dr. Ali Karimpour Nov 2014
Lecture 4
28
Topics to be covered include: State space solution.
State transition matrix. State transition equation.
Controllability and Pole placement by state feedback. Observability and estate estimation. Similarity transformation and canonical forms. Eigenvalues of the matrix A and poles of transfer function. Controllability and observability from block diagram.
Internal Description Model. (SS Description)
Dr. Ali Karimpour Nov 2014
Lecture 4
29
State Space ModelsGeneral form of state space model
LTI systems
مدل فضاي حالت
If initial condition and input are defined, then x(t), y(t) ?
و x(t)اگر شرائط اولیه و ورودي مشخص باشد مقدار چند است؟ y(t)
Start with homogeneous equation .با معادله همگن شروع کنیدAxx 00 )( xtx
Dr. Ali Karimpour Nov 2014
Lecture 4
30
State transition matrix ماتریس انتقال حالت
Axx )0()0( 0 xtx
)0()()( xttx
matrixtransitionstateis:)(t ماتریس انتقال حالت
)()()( 00 txtttx If t0 is not zero
Dr. Ali Karimpour Nov 2014
Lecture 4
31
Determination of state transition matrix
تعیین ماتریس انتقال حالتAxx 00 )( xtx
)()( 0 sAxxssx 01)()( xAsIsx
011 )()( xAsILtx
11 )()( AsILt
0)()( xttx
Method I
Dr. Ali Karimpour Nov 2014
Lecture 4
32
Determination of state transition matrix
تعیین ماتریس انتقال حالتAxx 00 )( xtx
0)( xetx At 0)()( xttx
Atet )(
...3.2.1
12.1
1 3322 tAtAAtIeAt
Method II
Dr. Ali Karimpour Nov 2014
Lecture 4
33
Determination of state transition matrix
تعیین ماتریس انتقال حالت
)0()0(
)()()()(
)()(
2
1
2221
1211
2
1
xx
tttt
txtx
)()(
01
)()()()(
)()(
21
11
2221
1211
2
1
tt
tttt
txtx
)()(
10
)()()()(
)()(
22
12
2221
1211
2
1
tt
tttt
txtx
So ith column of Φ is the response of system to
0..1..00 ith
position
Method III
Dr. Ali Karimpour Nov 2014
Lecture 4
34
Example 8: Find state transition matrix and x(t) of following system.
xx
3012
12
)0(x
011 )()( xAsILtx
1111
3012
)()(s
sLAsILt
310
)3)(2(1
21
1
s
sssL
t
ttt
eeee
t3
322
0)(
t
tt
t
ttt
eee
eeee
txtx
3
32
3
322
2
1 312
0)()(
Determination of state transition matrix
Dr. Ali Karimpour Nov 2014
Lecture 4
35
1- Φ(0) = I
LTIخواص ماتریس انتقال حالت در سیستمهاي
2- Φ-1(t)= Φ(-t)
3- Φ(t2-t1)Φ(t1-t0) = Φ(t2-t0)
5- Φk(t) = Φ(kt)
Property of state transition matrix for LTI systems
4- Φ(t2-t1) = Φ(t2) Φ-1(t1)
Dr. Ali Karimpour Nov 2014
Lecture 4
36
State transition matrix for time varying systems
ماتریس انتقال حالت براي سیستمهاي متغیر با زمانx
tx
000
01
)0()0(
)0(2
1
xx
x
25.0
1)(
ttx
10
)0()0(
)0(2
1
xx
x
10
)(tx
15.001
)( 2tt
)()(0)(
12
1
ttxtxtx
Dr. Ali Karimpour Nov 2014
Lecture 4
37
State transition equation معادله انتقال حالت
BuAxx 00 )( xtx
)()()( 0 sBusAxxssx
)()()()( 10
1 sBuAsIxAsIsx
dButxttxt
)()()()(00 Convolution
integral
State transition equation
Dr. Ali Karimpour Nov 2014
Lecture 4
38
Example 9: Find state transition equation for unit step
uxx
10
3012
21
)0(x
011 )()( xAsILtx
1111
3012
)()(s
sLAsILt
310
)3)(2(1
21
1
s
sssL
t
ttt
eeee
t3
322
0)(
State transition equation
Dr. Ali Karimpour Nov 2014
Lecture 4
39
uxx
10
3012
21
)0(x
t
ttt
eeee
t3
322
0)(
dbutxttxt
)()()()(00
21
0)(
3
322
t
ttt
eeee
tx
due
eeet
t
ttt
)(10
00 )(3
)(3)(2)(2
t
tt
eee
tx3
32
223
)(
de
eet
t
tt
0 )(3
)(3)(2
.....
.....
State transition equation
These are eigenvalues of A.
Dr. Ali Karimpour Nov 2014
Lecture 4
40
Example 10: Find x1(s) and x2(s)
)()()0()()( 11 sBuAsIxAsIsx
uxx
10
3012
)0()0(
)0(2
1
xx
x
)(10
3012
)0(30
12)(
11
sus
sx
ss
sx
)(
31
)3)(2(1
)0()0(
310
)3)(2(1
21
)()(
2
1
2
1 su
s
ssxx
s
ssssxsx
)(3
1)0(3
1
)()3)(2(
1)0()3)(2(
1)0(2
1
2
21
sus
xs
suss
xss
xs
State transition equation
Dr. Ali Karimpour Nov 2014
Lecture 4
41
Example 11: Derive x1(s) by using state diagram.
uxx
10
3012
1x 1xs -1 s -1
)0(1x
1
-21x2x
2x 2xs -1 s -1
)0(2x
1u(s)
-3
)()3)(2(
1)0()3)(2(
1)0(2
1)( 211 suss
xss
xs
sx
State transition equation
Dr. Ali Karimpour Nov 2014
Lecture 4
42
Topics to be covered include: State space solution.
State transition matrix. State transition equation.
Controllability and Pole placement by state feedback. Observability and estate estimation. Similarity transformation and canonical forms. Eigenvalues of the matrix A and poles of transfer function. Controllability and observability from block diagram.
Internal Description Model. (SS Description)
Dr. Ali Karimpour Nov 2014
Lecture 4
43
Controllability
Definition 1:The state equation (I) or the pair (A,b) is said to be controllable if
for any initial state x0 and any final state x1, there exists an input that transfers
x0 to x1 in a finite time. Otherwise (I) or (A,b) is said to be uncontrollable
، ورودي اي x1 براي هرو x0را کنترل پذیر گویند اگر براي هر (A,b)یا زوج (I)معادالت : 1تعریفیا زوج (I)معادالت در غیر اینصورت . برساند x1 را در زمان محدود به x0وجود داشته باشد که
(A,b) را کنترل ناپذیر گویند
، ورودي اي 1x براي هرو 0x براي هرگویند اگر کنترل پذیررا (A,b)یا زوج (I)معادالت : 1تعریفیا زوج (I)معادالت در غیر اینصورت . برساند x1 به زمان محدودرا در x0وجود داشته باشد که
(A,b) ناپذیر گویندرا کنترل
)(IducxybuAxx
Dr. Ali Karimpour Nov 2014
Lecture 4
44
Controllability testتست کنترل پذیري
Theorem 1:The state equation (I) or the pair (A,b) is controllable if and only if
اگر فقط و اگر است پذیر کنترل (A,b) زوج یا (I) معادالت :1قضیه
)(IducxybuAxx
S= [b Ab A2b ….. An-1b], | S| ≠ 0
Dr. Ali Karimpour Nov 2014
Lecture 4
45
Example 12: Check the controllability of following system
xy
uxx
]011[100
6116100010
2561610100
][ 2bAAbbS
12561610100
S
This kind of system is controllable so it is called controllable canonical form.
فرم کانونی کنترل پذیراین نوع سیستم همواره کنترل پذیر است لذا به آن . گویند
Controllability test
Dr. Ali Karimpour Nov 2014
Lecture 4
46
Use of state feedback to control a system
استفاده از فیدبک حالت براي کنترل سیستم
ducxybuAxx
x x
c
A
u
y
b
d
k
State feedback ? kxru
vector1nconstantn 1
r
drxdkcybrxbkAx
)()(
0A-sI of roots
0bkA-sI of roots
Dr. Ali Karimpour Nov 2014
Lecture 4
47
Use of state feedback to control a system
ducxybuAxx
استفاده از فیدبک براي کنترل سیستم
Is it possible to assign the eigenvalues arbitrarily? drxdkcybrxbkAx
)()(
state feedbackkxru
)(I
Theorem: If the n-dimensional state equation (I) is controllable, then bystate feedback u=r-kx, where k is a 1×n real constant vector, theeigenvalues of A-bk can arbitrarily be assigned provided that complex
conjugate eigenvalues are assigned in pairs.
حالت فیدبک با آنگاه باشد، پذیر کنترل (I) بعدي- n حالت معادالت اگر :قضیهu=r-kx ، که k 1 ثابت بردار یک×n ویژه مقادیر توان می باشد، می A-bk بطور را
بصورت باید مختلط ویژه مقادیر که نمود توجه باید البته نمود تعیین دلخواه.شود انتخاب مزدوج
Dr. Ali Karimpour Nov 2014
Lecture 4
48
Example 13: Is it possible to assign the eigenvalues of following system on arbitrarily position؟
uxx
021
7101401
800
132028121601
2bAAbbS
21132028121601
S System is controllable
So It is possible to assign the eigenvalues on arbitrarily position.
.لذا می توان مقادیر ویژه را در مکانهاي دلخواه قرار داد
Use of state feedback to control a system
Dr. Ali Karimpour Nov 2014
Lecture 4
49
Example 14: Is it possible to assign eigenvalues of following system by state feedback.
So : s=2 is a fixed mode.
Methods of pole placement
Clearly no but why?
Dr. Ali Karimpour Nov 2014
Lecture 4
50
Controllability test for each mode
تست کنترل پذیري براي هر مود
ducxybuAxx
Eigenvalues of A
Aمقادیر ویژه 1
2 3
n..
bAIi |
For the controllability of a mode it is sufficient that the following matrix has full row rank
i
......در حالت مقادیر ویژه تکراري : نکته
Dr. Ali Karimpour Nov 2014
Lecture 4
51
Example 15: Check the controllability of each mode
xy
trxx
]11[
)(01
1012.
0
0021
SS It is not completely controllable
2,1)2)(1(10
12A-sI 21
sss
s
000111
:1 ofility Controllab 1
lecontrollabnot is1 1
010110
:2 ofility Controllab 2
rank row fullnot
rank row full lecontrollab is2 2
Controllability test for each mode
Dr. Ali Karimpour Nov 2014
Lecture 4
52
Example 16: Consider following system.
uxx
021
300020001
System is not controllableI) Is it possible assign the eigenvalues on an arbitrarily places?
I (می توان مقادیر ویژه را در مکانهاي دلخواه قرار داد؟ آیاClearly No
واضح است که خیرII) Is it possible to assign the eigenvalues on -3, -4, -2 ?
II (قرار داد؟ 4- ,3- ,2-می توان مقادیر ویژه را در آیا Clearly Yes
واضح است که بلهIII) Is it possible to the eigenvalues on -13, -2±3j ?
III (3±2- ,13-می توان مقادیر ویژه را در آیاj قرار داد؟ Clearly No
واضح است که خیرIv) Is it possible to the eigenvalues on -3, -2+3j, -2-6j ?
Iv (3+2- ,3-می توان مقادیر ویژه را در آیاj, -2-6j قرار داد؟ Clearly No
واضح است که خیر
Use of state feedback to control a system
Dr. Ali Karimpour Nov 2014
Lecture 4
53
Methods of pole placement
روشهاي تعیین محل قطبها
1- Direct method. 1- روش مستقیم
2- Use of similarity transformation. استفاده از تبدیالت همانندي -2
Dr. Ali Karimpour Nov 2014
Lecture 4
54
Direct method of pole placement
bkA-sI:Find
روش مستقیم تعیین محل قطبها
= Desired characteristic equation
Then determine k1, k2, … , kn from above equation
ducxybuAxx
xkkkru n ]...[Let 21
Polynomial ofDegree n
Polynomial ofDegree n
Dr. Ali Karimpour Nov 2014
Lecture 4
55
Example 17: Assign the poles on -2±j if it is possible.
uxx
11
1011
11
01S
01S
System is controllable
So it is possible to assign the poles on -2±j .
][11
1011
10 kkbkA
10
10
111kkkk
10
10
111
kskkks
bkAsI
110
2 1)( kskks
)2)(2(1)( 1102 jsjskskks 542 ss ]610[ k
Method I
Use of similarity transformation for pole placement
Dr. Ali Karimpour Nov 2014
Lecture 4
56
Topics to be covered include: State space solution.
State transition matrix. State transition equation.
Controllability and Pole placement by state feedback. Observability and estate estimation. Similarity transformation and canonical forms. Eigenvalues of the matrix A and poles of transfer function. Controllability and observability from block diagram.
Internal Description Model. (SS Description)
Dr. Ali Karimpour Nov 2014
Lecture 4
57
Observability
Definition 2:The state equation (I) or the pair (A,c) is said to be observable if
for any unknown initial state x0 , there exists a finite time t1 > 0 such that the
knowledge of the input u and the output y over [0,t1] suffices to determine
Uniquely the initial state x0. Other wise, the equation is unobservable.
x0 براي هر شرط اولیهگویند اگر رویت پذیررا (A,c)یا زوج (I)معادالت : 2تعریف ]t0,1[در بازه yو خروجی u، وجود داشته باشد که اطالعات ورودي 1t<0زمان محدود
.کافی باشد در غیر اینصورت سیستم را رویت ناپذیر گویند x0 منحصر بفردبراي محاسبه
)(IducxybuAxx
Dr. Ali Karimpour Nov 2014
Lecture 4
58
Observability test
Theorem 2:The state equation (I) or the pair (A,c) is observable if and only if
اگر فقط و اگر است پذیر رویت (A,c) زوج یا (I) معادالت :2قضیه
0,
.
.
1
2
V
cA
cAcAc
V
n
)(IducxybuAxx
Dr. Ali Karimpour Nov 2014
Lecture 4
59
Example 18: Check the observability of following system
xy
uxx
]011[100
6116100010
5116110011
2cAcAc
V
0)60(1)115(15116110011
VSo the system is not
observable.
پس سیستم رویت پذیر نیست
Observability test
Dr. Ali Karimpour Nov 2014
Lecture 4
60
Observability test for each mode (system with distinct eigenvalues)
)سیستمهاي داراي مودهاي مجزا(تست رویت پذیري براي هر مود
ducxybuAxx
Eigenvalues of A
Aمقادیر ویژه 1
2 3
n..
c
AIi
For the observability of a mode it is sufficient that the following matrix has full column rank
i
Dr. Ali Karimpour Nov 2014
Lecture 4
61
Use of state estimation to use in feedback loop
استفاده از تخمین زن حالت براي استفاده در مسیر فیدبک
x x
c
A
u
y
b
d
k
r
States are not available!
x Estate Estimator
x
Condition for xx ˆ
Dr. Ali Karimpour Nov 2014
Lecture 4
62
Use of state estimation to use in feedback loop
استفاده از تخمین زن حالت براي استفاده در مسیر فیدبک
x x
c
A
u
y
b
d
k
r
x Estate Estimator
x
)ˆ(ˆˆ xccxlbuxAx
)ˆ(ˆˆ xcylbuxAx
Dr. Ali Karimpour Nov 2014
Lecture 4
63
Use of state estimation to use in feedback loop
استفاده از تخمین زن حالت براي استفاده در مسیر فیدبک
)ˆ(ˆˆ xcylbuxAx
x x
c
A
u
y
b
d
k
r
x
x
A
b
c
l
Estimator
Dr. Ali Karimpour Nov 2014
Lecture 4
64
Topics to be covered include: State space solution.
State transition matrix. State transition equation.
Controllability and Pole placement by state feedback. Observability and estate estimation. Similarity transformation and canonical forms. Eigenvalues of the matrix A and poles of transfer function. Controllability and observability from block diagram.
Internal Description Model. (SS Description)
Dr. Ali Karimpour Nov 2014
Lecture 4
6565
Change of variables (Similarity transformation)
323
212
311
2xxwxxwxxw
Pxxw
110021101
wwPx
1
1
110021101
ducxybuAxx
duwcPyPbuwPAPw
1
1
udwcy
ubwAwˆˆ
ˆˆ
A b c d
ducxybuAxx
113113 33
isdiscisbisA
Let
wP 1
wAP 1 bu
y wcP 1 du
Dr. Ali Karimpour Nov 2014
Lecture 4
6666
]06.04.0[ˆ 1 cPc
Example 19: A similarity transformation example
xy
uxx
]122[43
6
661221463
30
9.4ˆ Pbb
0ˆ dd
wy
uww
]06.04.0[30
9.4
400010002
21
42 13
Eigenvalues of AAمقادیر ویژه
21
42 13
Eigenvalues of Aمقادیر ویژه
A
New system is very simpler than older
400010002
ˆ 1PAPA
21-1-1-1121-2
1P
Eigenvector corresponding to -2
Change of variables (Similarity transformation)
Dr. Ali Karimpour Nov 2014
Lecture 4
67
Similarity transformation properties
ddcPc
PbbPAPA
ˆˆ
ˆˆ1
1ducxybuAxx
Pxw
ation transformSimilarity
udwcy
ubwAwˆˆ
ˆˆ
1- It can lead to a simpler system. امکان ساده سازي سیستم -12- It doesn’t change the eigenvalues. عدم تغییر مقادیر ویژه -23- Similar transfer functions. تابع انتقال یکسان - 3
5- It doesn’t change observability. عدم تغییر رویت پذیري - 54- It doesn’t change controllability. عدم تغییر کنترل پذیري -4
Dr. Ali Karimpour Nov 2014
Lecture 4
68
Effect of similarity transformation on eigenvalues of system
تاثیر تبدیالت همانندي بر مقادیر ویژه
ddcPc
PbbPAPA
Pxw
ˆˆ
ˆˆ1
1
ducxybuAxx
udwcy
ubwAwˆˆ
ˆˆ
Similarity transformation doesn’t change the eigenvalues.تبدیل همانندي مقادیر ویژه را تغییر نمی دهد
0 AsI 0ˆ AsI
AsI ˆ 11 PAPsPP 1)( PAsIP 1 PAsIP
AsI
Dr. Ali Karimpour Nov 2014
Lecture 4
69
Effect of similarity transformation on controllability
تاثیر تبدیالت همانندي بر کنترل پذیري
ddcPc
PbbPAPA
ˆˆ
ˆˆ1
1
ducxybuAxx
udwcy
ubwAwˆˆ
ˆˆ
Similarity transformation doesn’t change the controllability.تبدیل همانندي کنترل پذیري را تغییر نمی دهد
bAbAAbbS n 12 ...
bAbAbAbS n ˆˆ...ˆˆˆˆˆˆ
12
PbPPAPbPPAPbPAPPb n 11121 ... bAbAbAbS n ˆˆ...ˆˆˆˆˆˆ 12
bPAbPAPAbPb n 12 ... bAbAAbbP n 12 ...
Matrix ility Controllab
PS
SPPSS ˆ
Dr. Ali Karimpour Nov 2014
Lecture 4
70
Effect of similarity transformation on observability
تاثیر تبدیالت همانندي بر رویت پذیري
ddcPc
PbbPAPA
ˆˆ
ˆˆ1
1
ducxybuAxx
udwcy
ubwAwˆˆ
ˆˆ
Similarity transformation doesn’t change the observability.تبدیل همانندي رویت پذیري را تغییر نمی دهد
1
.
.
ncA
cAc
VMatrix
ity Observabil
1ˆˆ
.
.
ˆˆ
ˆ
ˆ
nAc
Ac
c
V
With the same proof as pervious slide
Dr. Ali Karimpour Nov 2014
Lecture 4
71
Important canonical forms
فرمهاي کانونی مهم
1- Controllable canonical form. فرم کانونی کنترل پذیر -1
2- Observable canonical form. فرم کانونی رویت پذیر -2
3- Jordan canonical form. فرم کانونی جردن -3
Dr. Ali Karimpour Nov 2014
Lecture 4
72
Controllable canonical form
فرم کانونی کنترل پذیر
xy
uxx
]...[10.00
...1...000...........0...1000...010
...1
...10...........
...0001...000
S 0 S
xy
uxx
]021[100
7148100010
814712)( 23
sssssG
Dr. Ali Karimpour Nov 2014
Lecture 4
73
Transforms to controllable canonical form
uxx
11
1011
تبدیل به فرم کانونی کنترل پذیر
0110ˆ 1PAPA
10ˆ Pbb
11
01S
01SSystem is
controllable
1
...nqA
qAq
P q is last rowof S-1
0111
P
uww
10
0110
Controllable canonical form
Dr. Ali Karimpour Nov 2014
Lecture 4
74
Observable canonical form
فرم کانونی رویت پذیر
xy
uxx
]1...000[
.1000
..............010...001...000
...1...........
1...0010...00
V 0 V
xy
uxx
]100[021
7101401
800
814712)( 23
sssssG
Dr. Ali Karimpour Nov 2014
Lecture 4
75
Transforms to observable canonical form
xy
uxx
]10[11
1221
تبدیل به فرم کانونی رویت پذیر
13ˆ Pbb
1210
VSystem is observable
qAAqqP n 11 ... q is last columnof V-1
1012
P
wy
uww
]10[13
2150
Observable
canonical form
2150ˆ 1PAPA
0ˆ
10ˆ 1
dd
cPc
Dr. Ali Karimpour Nov 2014
Lecture 4
76
Jordan canonical form for systems with distinct eigenvalues
فرم کانونی جردن براي سیستم با مقادیر ویژه مجزا
xccccy
u
b
bb
xx
n
nn
]...[
.
.
...000......................0...000...00
321
2
1
2
1
xy
uxx
]33.05.017.0[137
100020004
Checking the controllability and observability is very simple
All modes are controllable and observable
تمام مودها کنترل پذیرو رویت پذیر است
4)7)(17.0()(
s
sg2
)3()5.0(
s 1)1()33.0(
s
Dr. Ali Karimpour Nov 2014
Lecture 4
7777
]06.04.0[ˆ 1 cPc
Remember: A similarity transformation example
xy
uxx
]122[43
6
661221463
30
9.4ˆ Pbb
0ˆ dd
wy
uww
]06.04.0[30
9.4
400010002
21
42 13
Eigenvalues of AAمقادیر ویژه
400010002
ˆ 1PAPA
21-1-1-1121-2
1P
Eigenvector corresponding to -2
Change of variables (Similarity transformation)
2)9.4)(4.0()(
s
sg1
)0()6.0(
s 4
)3()0(
s
296.1)(
s
sg
؟ Aمقادیر ویژه و قطبهاي تابع انتقالارتباط بین
Dr. Ali Karimpour Nov 2014
Lecture 4
78
Eigenvalues of matrix A and poles of transfer function
xy
uxx
]011[100
6116100010
31
22 13
Eigenvalues of AAمقادیر ویژه 31 p
22 p
Poles of systemقطبهاي سیستم
What happened to -1 ?
)3)(2(1)(
sssg
[num,den]=ss2tf(A,b,c,0), , g=minreal(g)g=tf(num,den)
Dr. Ali Karimpour Nov 2014
Lecture 4
79
Extended Jordan canonical form
فرم کانونی جردن تعمیم یافته
xcccy
u
b
bb
xx
r
rr
]...[
....000
...........0...000...000...01
21
2
1
2
1
1
iff econrollabl are and 11 0 1 b
iff econrollabl is 2 0 2 b
0 rbiff econrollabl is r
.
...
iff observable are and 11 0 1 c
iff observable is 2 0 2 c
0 rciff observable is r
.
...
Dr. Ali Karimpour Nov 2014
Lecture 4
80
Transfer function of controllable and observable systems.
تابع انتقال سیستمهاي کنترل پذیر و رویت پذیر
nnbeALetducxybuAxx
dbAsIcsG 1)()(
Suppose the system is controllable and observableفرض کنید سیستم کنترل پذیر و رویت پذیر باشد
Suppose the system is not controllable or is not observableنیستفرض کنید سیستم کنترل پذیر یا رویت پذیر
Dr. Ali Karimpour Nov 2014
Lecture 4
81
Example 20: Transfer function of controllable and observable systems.
xy
uxx
]111[011
100021321
xy
uxx
]201[110
300010011
32)(
s
sG
375132)( 23
2
sss
sssGSystem is
controllable and
observable
System is not both controllable and
observable but 3 is controllable and
observableBut we can not know much more information about the system by this method
.اما نمی توان اطالعات بیشتري از این روش بدست آورد
Dr. Ali Karimpour Nov 2014
Lecture 4
82
Topics to be covered include: State space solution.
State transition matrix. State transition equation.
Controllability and Pole placement by state feedback. Observability and estate estimation. Similarity transformation and canonical forms. Eigenvalues of the matrix A and poles of transfer function. Controllability and observability from block diagram.
Internal Description Model. (SS Description)
Dr. Ali Karimpour Nov 2014
Lecture 4
83
Controllability and observability from block diagram and state diagram
کنترل پذیري و رویت پذیري از طریق بلوك دیاگرام و دیاگرام حالت1- Find the numbers of s-1in the block diagram or state diagram.
را در بلوك دیاگرام و یا دیاگرام حالت بدست آورید s-1ابتدا تعداد - 12 - Find the transfer function by Mason’s rule
تابع انتقال را با توجه به قانون گین میسون بدست آورید - 2
3 – If the numbers of s-1in the block diagram or state diagram is equal to degree of transfer function then the system is controllable and observable.
برابر در بلوك دیاگرام و یا معادالت حالت با درجه تابع انتقال s-1اگر تعداد - 3.است کنترل پذیر و رویت پذیرآنگاه سیستم با شد
Dr. Ali Karimpour Nov 2014
Lecture 4
84
Example 21: Check the controllability and observability of following function.
TF
451
)()()( 2
sssrscsG
Mason’s rule
The numbers of s-1in the state diagram is 2 and the degree of transfer function is also 2 so the system is controllable and observable.
است لذا 2بوده و درجه تابع انتقال نیز 2در دیاگرام حالت برابر s-1تعداد .است کنترل پذیر و رویت پذیرسیستم
Controllability and observability from block diagram and state diagram
Dr. Ali Karimpour Nov 2014
Lecture 4
85
Exercises
3334)( 23
2
ssS
ssG4-1 Find the poles and zeros of following system.
4-2 Is it possible to apply a nonzero input to the following systemfor t>0, but the output be zero for t>0? Show it. uuyyy 265
4-3 Find the step response of following system.333
4)( 23
2
sss
ssG
4-4 Find the step response of following system for a=1,3,6 and 9.
222)( 2
ssassG
)(sin1)( 2 tuttetty t 4-5 Find the poles of Y(s)
)1(1)(
ss
sY4-6 Check F.V.T and I.V.T for following system.
Dr. Ali Karimpour Nov 2014
Lecture 4
86
Exercises
4-7) Find the state diagram of the following system
uxx
010
141230312
4-8) Find the transfer function of the following system.a) By use of the formula between the two representation.b) By use of state diagram
xy
uxx
]021[010
141230312
4-9) Find y(t) for initial condition [1 3 -1]T and the unit step as input.
xy
uxx
]021[010
141230312
Dr. Ali Karimpour Nov 2014
Lecture 4
87
Exercises
4-10) The state transition matrix and state transition equation of following system.
)(21
)(4031
)( trtXtX
t
ttt
eeee
s
sssLs
sLAsILt
4
421
1111
04
10
453
11
4031
)()(
dtReeee
Xeeee
tXt
t
ttt
t
ttt
0 )(4
)(4)()(
4
4
)(21
0)0(
0)(
Answer is:
4-11) Suppose x1(0)=1, x2(0)=3 and x3(0)=2 and r(t)=u(t) . Find x(t) and c(t) for t≥0
)(]001[)()(011
)(300122041
)( tXtctrtXtX
Dr. Ali Karimpour Nov 2014
Lecture 4
88
Exercises
4-13) a)Find the eigenvalues of following system.b) Find the pole of transfer function.c) Are the eigenvalues controllable and observable?
xy
uxx
]021[010
141230312
4-14) Check the controllability and observability of every mode in the following system.
xy
uxx
]122[43
6
661221463
4-12) Find the modes of following system.uxx
010
141230312
Dr. Ali Karimpour Nov 2014
Lecture 4
89
Exercises4-15) Check the controllability and observability of every mode in the following system.
wy
uww
]011[102
400010002
4-16) Examine the controllability and observability of each mode of the following system.
)()(]101[)()(101
)(400320001
)( trtXtctrtXtX
Answer: All modes are controllable. -1 and -4 are observable but -2 is not observable.4-17) The eigenvalues of a system are -3,4,-5 and the poles of its transfer function are -3 and -5.(Midterm spring 2008)a) What is your idea about the observability of each eigenvalue.b) What is your idea about the controllability of each eigenvalue.
Dr. Ali Karimpour Nov 2014
Lecture 4
90
Exercises
4-19) Check the controllability and observability of every mode in the following system by direct inspection (without calculation). wy
uww
]011[102
400010002
S-1
-2
11 -3
S-1 -21
S-1
-1 31
r(s) c(s)4-20) Check the controllability and observability of the following system.
4-18) Consider the following system.a) Find the controllable canonical form. b) Find the observable canonical form.
xy
uxx
]122[43
6
661221463
Dr. Ali Karimpour Nov 2014
Lecture 4
91
4-21) Check the controllability and observability of the following system by transfer function method.
21s
s13
+
+
++
Exercises
4-22) Find a state space realization for following system such that it is neither observable nor controllable.(Final 1391)
12)(
sssg
4-23) Is it possible to assign the eigenvaluesof following system on arbitrarily places?
xy
uxx
]021[010
141230312
Dr. Ali Karimpour Nov 2014
Lecture 4
92
xy
uxx
]122[43
6
661221463
4-24) Is it possible to assign the eigenvalues of following system on arbitrarily places?
wy
uww
]011[102
400010002
4-25) Use a state feedback to assign the eigenvalues of following system on -4, -1 , -6?b) Use a state feedback to assign the eigenvalues of following system on -4, -2 , -6?c) Use a state feedback to assign the eigenvalues of following system on -1, -2 ±6j?
Exercises
Dr. Ali Karimpour Nov 2014
Lecture 4
93
Exercises4-26) Is it possible to assign the eigenvalues of following system on arbitrarily places. If yes, put them on -2, -3 and -4.
uxx
021
310020001
4-27) Is it possible to assign the eigenvalues of following system on arbitrarily places (Final 1391)? If yes, put them on -2 and -4.
xy
uxx
]11[5
1123510
Dr. Ali Karimpour Nov 2014
Lecture 4
94
Use of similarity transformation for pole placement
استفاده از تبدیالت همانندي در تعیین محل قطبها
buAxx Pxw uw
aaaa
w
n
10..00
...1...000
...............................................0...1000...010
1210
?ˆˆˆ kbA
11221100ˆ...ˆˆˆ
1...000...............................................................................0...1000...010
nn kakakaka
012
21
1 ... :is system ofequation sticCharacteri asasasas nn
nn
n
012
21
1 ... :isequation sticcharacteri Desired bsbsbsbs nn
nn
n
2b 1 nb1b0b
Dr. Ali Karimpour Nov 2014
Lecture 4
95
Use of similarity transformation for pole placement استفاده از تبدیالت همانندي در تعیین محل قطبها
11221100ˆ...ˆˆˆ
1...000...............................................................................
0...1000...010
ˆˆˆ
nn kakakaka
kbA
2b 1 nb1b0b
11221100 ....ˆ nn ababababk
wkru ˆ Pkk ˆPxkr ˆ kxr
x x
c
A
u
y
b
d
Pw
k
k
r
Dr. Ali Karimpour Nov 2014
Lecture 4
96
01S
System is controllable
So it is possible to assign the poles on -2±j .
Method II
qAq
P 1]10[ Sq
0111
P
]46[]04)1(5[ˆ1100 ababk
System characteristic equation: s2+0s-1Desired characteristic equation: (s+2+j)(s+2-j)=s2+4s+5
]610[ˆ Pkk
uxx
11
1011
11
01S
place(a,b,[-2+i -2-i])
Use of similarity transformation for pole placement
Example 22: Assign the poles on -2±j if it is possible.