Controllability and

Observability

Dr. S.2.06.2022
AYASUN 2

Controllability

1
Dr. S.2.06.2022
AYASUN 3

Controllability

Dr. S.2.06.2022
AYASUN 4

Controllability

2
Dr. S.2.06.2022
AYASUN 5

Motivation Example
 x1   2 1   x1  1
 x    0  1  x   0u (t )
 2   2   
x 
y  1 0 1 
 x2 
x2 (0) x1 (0)
uncontrollable s s

u x2 s 1 x2 1 x1 s 1 x1 1 y

1 2
1
controllable

Dr. S.2.06.2022
AYASUN 6

Controllability Test

3
Dr. S.2.06.2022
AYASUN 7

Proof of controllability matrix

xk 1  Axk  Bu k
xk  2  Axk 1  Bu k 1
xk  2  A( Ax k  Bu k )  Bu k 1  A2 xk  ABu k  Bu k 1
xk  n  An xk  An 1 Bu k  An  2 Bu k 1    ABu k  ( n  2 )  Bu k  ( n 1)
xk  n  An xk  An 1 Bu k  An  2 Bu k 1    ABu k  ( n  2 )  Bu k  ( n 1)
 uk 
  

xk  n  An xk  An 1 B  AB B  
uk  ( n  2 ) 
 
 uk  ( n 1) 
Initial condition

Dr. S.2.06.2022
AYASUN 8

Example 1: System 1

u 1 s+2 y
s+2 s+3
x2 x1
s2
x̂1 (s)  x̂ 2 (s); s  3x̂1(s)  s  2 x̂ 2 (s); x 1  3x1  x 2  2x 2
s3
1
x̂ 2 (s)  û (s); s  2 x̂ 2 (s)  û(s); x 2  2x 2  u; x 1  3x1  u
s2
 3 0  1 1  3
x    x    u; y  x1  1 0x C  B AB   
 0  2 1 1  2

Rank C = 2 = n, therefore the system is controllable

4
Dr. S.2.06.2022
AYASUN 9

Example 2: System 2
u s+2 1 y
s+3 v s+2 x1
1
x̂1 (s)  v̂ (s); s  2 x̂1 (s)  v̂ (s); x 1  2x1  v
s2
s2 û (s) 1 s  3x̂ 2 (s)  û(s);
v̂ (s)  û (s)  (s  2); x̂ 2 (s)  û(s);
s3 s3 x s3 x 2  3x 2  u;
2

v̂ (s)  s  2 x̂ 2 (s); v  x 2  2x 2   x 2  u; x 1  2x1  x 2  u;

 2  1 1 1  3
x    x    u; y  x1  1 0x; C  B AB   
 0  3 1 1  3
Rank C = 1 < 2, therefore the system is not controllable

Dr. S.2.06.2022
AYASUN10

Controllability and Algebraic Equivalence

5
Dr. S.2.06.2022
AYASUN11

Example 3 (System 1 continued)

u 1 s+2 y
s+2 s+3
 3 0  1
x    x    u; y  x1  1 0x
 0  2 1
• What is the new representation?
1  3 = Q;  2 3
C  B AB    P  Q 1   
1  2   1 1
 2 3   3 9 0  6
A  PAP 1     
  1 1  2 4 1  5
 2 3 1 1
B  PB        ; C  CP 1  1  3
  1 1 1 0

Dr. S.2.06.2022
AYASUN12

Example 4 (System 2 continued)

u s+2 1 y
s+3 v s+2 x1

 2  1 1
x    x    u; y  x1  1 0x
 0  3 1
1  3
C  B AB   
1  3
• What can be done?
– Use b as one vector, and find another vector to
form a basis
1 0  1 0
Q ; P  Q 1   
1 1   1 1

6
 Example 4 (System 2 continued)
Dr. S.2.06.2022
AYASUN13

 1 0  3  1  3  1
A  PAP 1     
 1 1  3  3  0  2
 1 0 1 1
B  PB        ; C  CP 1  1 0
 1 1 1 0
  3  1 1
x    x    u; y  1 0x
 0  2 0
• What can you see?
u + 1 y
_
- s+3 x1
1
_
s+2 x2 ~ Not controllable

Dr. S.2.06.2022
AYASUN14

Example 5: Frobenious Form

 0 1 0  0

x 0 0 
1 x  0  u
   
 18 27 10  1 

ctrb Compute the controllability matrix.

CO = ctrb(A,B) returns the controllability matrix [B AB A^2B ...].

CO = ctrb(SYS) returns the controllability matrix of the

state-space model SYS with realization (A,B,C,D). This is
equivalent to ctrb(sys.a,sys.b).
A=[0 1 0;0 0 1;-18 -27 -10], b=[0;0;1], C=ctrb(A,b)

C =[ 0 0 1; 0 1 -10; 1 -10 73]

rank(C)=3

7
Dr. S.2.06.2022
AYASUN15

Example 6: Controller Form

 9 24 20  1 

x 1 0 0 x  0  u

   
 0 1 0  0 
>> A=[-9 -24 -20;1 0 0;0 1 0]

A=
-9 -24 -20
1 0 0
0 1 0
>> b=[1;0;0]
b=
1
0
0
>> C=ctrb(A,b)
C =1 -9 57
0 1 -9
0 0 1
>> rank(C) = 3

Dr. S.2.06.2022
AYASUN16
Popov-Belovich-Hautus (PBH) test) for
Controllability

  3  1 1
x    x    u; y  1 0x
 0  2 0
Eigenvalues are: 1= -3 and 2 = -2
  3 1    3 1 1
 I  A   0   2 , I  A b   0   2 0
   
  3 1 1 0 1 1 
ranh  1I  A b   rank  1   rank 0 1 0   2
 0 1  2 0   
controllable

1 1 1  1 1 1 
rank  2 I  A b   rank    rank   1
0 0 0  0 0 0 
Uncontrollable

8
Dr. S.2.06.2022
AYASUN17

Example 7

•Aşağıda verilen LZD sistem için

2 1 1 1 
x(t )   5 3 6  x(t )  0  u (t )
 
 5 1 4  0
y (t )  1 0 1 x(t )
• Özdeğerleri bulunuz.
• Sistem tamamı ile kontrol edilebilir mi?
• PBH rank testini kullanarak hangi modların (özdeğerlerin)
kontrol edilebilir ya da edilemez olduğunu belirleyiniz.
• Kontrol edilebilirlik matrisini kullanarak sistemi kontrol
edilebilir ve kontrol edilemez alt sistemlere ayırınız.

Dr. S.2.06.2022
AYASUN18

Example 7

9
Dr. S.2.06.2022
AYASUN19

Example 7

Dr. S.2.06.2022
AYASUN20

Example 7

10
 21
Canonical Decomposition

Theorem If (C) = nc < n, then  an equivalent

transformation s.t.
 x c   Ac A12   x c   Bc  x 
 x    0  u; y  Cc Cc   c   Du
 c  A c   x c   0  x c 

The subsystem x c  Ac x c  Bc u; y  Cc x c  Du
is controllable, and has the same transfer function as
the original system
D
+
Controllable +
u Subsystem y
+
Uncontrollable
Subsystem

Dr. S.2.06.2022
AYASUN22

Canonical Decomposition

11
Dr. S.2.06.2022
AYASUN23
Canonical Decomposition

Dr. S.2.06.2022
AYASUN24
Example 8

12
Dr. S.2.06.2022
AYASUN25

Example 8

Dr. S.2.06.2022
AYASUN26

Example 9

13
Dr. S.2.06.2022
AYASUN27

Example 9

Dr. S.2.06.2022
AYASUN28

Example 9

14
Dr. S.2.06.2022
AYASUN29

Example 9

Dr. S.2.06.2022
AYASUN30

Example 9

15
Dr. S.2.06.2022
AYASUN31

Example 9

Dr. S.2.06.2022
AYASUN32

Example 9

16
Dr. S.2.06.2022
AYASUN33

Example 9

Dr. S.2.06.2022
AYASUN34

Example 9

17
Dr. S.2.06.2022
AYASUN35

Example 9

Observability

18
Dr. S.2.06.2022
AYASUN37

Observability

Dr. S.2.06.2022
AYASUN38

Observability

19
 Motivation: Example

 x1   2 0   x1  3
 x    0  1  x   1u (t )
 2   2   
x 
y  1 0 1 
 x2  x2 (0) x1 (0)
s s

u 1 x2 s 1 x2 x1 s 1 x1 1 y

1 2
3 observable
unobservable

39

Dr. S.2.06.2022
AYASUN40

Observability

20
Dr. S.2.06.2022
AYASUN41

Observability

Dr. S.2.06.2022
AYASUN42

Observability Test

21
 43

Proof of Observability Matrix

xk 1  Ax k  Bu k
yk  Cx k  Du k  (1)
yk 1  Cx k 1  Du k 1
yk 1  C ( Axk  Bu k )  Du k 1  CAx k  CBu k  Du k 1  (2)
yk  n 1  CA n 1 xk  CA n  2 Bu k  CA n 3 Bu k 1    CBu k  ( n  2 )  Du k  ( n 1)  (n)
 C 
 CA 
(1), (2), (n)   x
   k
 n 1 
CA 

 yk  Du k yk 1  CBu k  Du k 1   CABu k  ( n 3)  CBu k  ( n  2 )  Du k  ( n 1) 

Inputs & outputs

Dr. S.2.06.2022
AYASUN44

Duality Controllability-Observability

22
Dr. S.2.06.2022
AYASUN45

Example 1

Dr. S.2.06.2022
AYASUN46

Example 1

23
Dr. S.2.06.2022
AYASUN47

Example 1

Dr. S.2.06.2022
AYASUN48

Example 1

24
Dr. S.2.06.2022
AYASUN49

Example 1

Example 2: Is it observable?

u s+2 1 y
s+3 s+2
 2  1 1
x    x    u; y  x1  1 0x
 0  3 1
C 1 0
O  
CA   2  1
– Rank of O is 2 ~ The system is observable
– Use rows of O as a basis:
1  2  1  2
Q ; P  Q1   
0  1  0  1 

25
 Example 2: Is it observable?

1  2  2 5  2  1
A  PAP 1     
0  1  0 3  0  3

1  2 1  1
B  PB        ;
0  1 1  1

C  CP 1  1  2

 2  1  1
x    x   1 u;
 0  3  

y  1  2x

~ Both modes (x1 andx2) contribute to y

52
Example 3: Is it observable?

u 1 s+2 y
s+2 s+3
 3 0  1  C   1 0
x    x    u; y  x1  1 0x; O  
 0  2 1 CA  3 0

• Rank is 1, and the system is not observable

• Through similar transformation, we have
1
u s+3 y
1
s+2
~x2 does not contribute to anything

26
53 S. AYASUN
Dr.

Example 4: Observable Canonical Form

 9 1 0  0 
x  24 0 1 x  1  u
  y  1 0 0 x
   
 20 0 0   3

obsv Compute the observability matrix.

OB = obsv(A,C) returns the observability matrix [C; CA; CA^2 ...]

CO = obsv(SYS) returns the observability matrix of the

state-space model SYS with realization (A,B,C,D). This is
equivalent to obsv(sys.a,sys.c).

>> A=[-9 1 0;-24 0 1;-20 0 0] C=[1 0 0]

ob=obsv(A,C) ob =[ 1 0 0; -9 1 0;57 -9 1]
>> rank(ob)
ans =

54 S. AYASUN
Dr.
2.06.2022

Example 4: Observable Form

0 0 20   3
 
x  1 0 24 x  1  u y  0 0 1 x
   
0 1 9  0 
>> A=[0 0 -20;1 0 -24;0 1 -9];
>> C=[0 0 1];
>> ob=obsv(A,C)

ob =[0 0 1; 0 1 -9;1 -9 57]

>> rank(ob)

ans =

27
Dr. S.2.06.2022
AYASUN55
The Popov-Belovich-Hautus (PBH) Test for
Observability

The n-dimensional pair (A, C) is observable if and only if the matrix

 λI  A 
 C 
 
has full column rank at every eigenvalue, , of A.

 3 0  1
x  x    u; y  1 0 x
 0 2  1

Eigenvalues are: 1= -3 and 2 = -2

  3 0 
  3 0    I  A 
 I  A   0   2 , ob   C    0   2
     1 0 


Dr. S.2.06.2022
AYASUN56
Canonical Decomposition

28
Dr. S.2.06.2022
AYASUN57
Canonical Decomposition

Dr. S.2.06.2022
AYASUN58
Example 5

29
Dr. S.2.06.2022
AYASUN59
Example 5

Dr. S.2.06.2022
AYASUN60
Canonical Decomposition: Duality

30
Dr. S.2.06.2022
AYASUN61
Canonical Decomposition: Duality

Dr. S.2.06.2022
AYASUN62

Example 6

31
Dr. S.2.06.2022
AYASUN63

Example 6

Dr. S.2.06.2022
AYASUN64

Example 6

32
Dr. S.2.06.2022
AYASUN65

Example 6

Dr. S.2.06.2022
AYASUN66

Example 6

33
Dr. S.2.06.2022
AYASUN67

Example 6

Dr. S.2.06.2022
AYASUN68

Example 6

34
Dr. S.2.06.2022
AYASUN69

Example 6

Dr. S.2.06.2022
AYASUN70

Example 6

35
Dr. S.2.06.2022
AYASUN71

Example 6

Dr. S.2.06.2022
AYASUN72

Example 6

36
Dr. S.2.06.2022
AYASUN73

Example 6

Dr. S.2.06.2022
AYASUN74

Example 6

37
Dr. S.2.06.2022
AYASUN75

Example 6

Dr. S.2.06.2022
AYASUN76

Example 6

38
Dr. S.2.06.2022
AYASUN77

Summary

Dr. S.2.06.2022
AYASUN78

Summary

39
Dr. S.2.06.2022
AYASUN79

Summary: Canonical Decomposition

Dr. S.2.06.2022
AYASUN80
Summary: Controllable/Uncontrollable Decomposition

40
Dr. S.2.06.2022
AYASUN81
Summary: Controllable/Uncontrollable Decomposition

Dr. S.2.06.2022
AYASUN82
Summary: Observable/Unobservable Decomposition

41
Dr. S.2.06.2022
AYASUN83
Summary: Observable/Unobservable Decomposition

Dr. S.2.06.2022
AYASUN84

Summary: Kalman Decomposition

42
Dr. S.2.06.2022
AYASUN85

Summary: Kalman Decomposition

Dr. S.2.06.2022
AYASUN86

Example 7

43
Dr. S.2.06.2022
AYASUN87

Example 7

Dr. S.2.06.2022
AYASUN88

Example 7

44
Dr. S.2.06.2022
AYASUN89

Example 7

Dr. S.2.06.2022
AYASUN90

Example 7

45
Dr. S.2.06.2022
AYASUN91

Example 7

Dr. S.2.06.2022
AYASUN92

Example 8

46
Dr. S.2.06.2022
AYASUN93

Example 8

Dr. S.2.06.2022
AYASUN94

Example 8

47
Dr. S.2.06.2022
AYASUN95

Example 8

48