Adama Science and Technology

University

Chapter 5
Controllability and Observability of a System

ASTU
School of EEC
5.1 Plant and Feedback Control
Consider Output feedback control
d
r+ e Controller u + + Plant y
- Gc(s) Gp(s)
+ n
+
Only output signal is used for feedback
5.1 Plant and Feedback Control
State feedback control
d
r Feedforward + u + + y
Plant
gain matrix -
Feedback x
gain matrix

Plant
x  Ax  Bu, x (t0 )  x0
y  Cx  Du
State variables are used for feedback.
5.2 Controllable/Uncontrollable
5.3 Controllability
Controllability concept
Situations:
 We want to design a feedback system with the
desired performance.
 We want to control unstable systems.

Questions:
 Is it possible to move the initial state x(t0) to the
desired state x(t1) after a finite time by applying
the appropriate input u(t)? YES.
5.3 Controllability
Observability concept
Situations:
 Some variables are difficult to measure for
feedback.
 Some sensors are too expensive. So they can not
be used.
Questions:
 Is it possible to know the initial state x(t0) or the
state x(t) from both the input u(t) and the output
y(t) observed for a finite time? YES.
5.3 Controllability
Why do we check it?
Necessary to know the existence
of uncontrollable state variables
because the feedback of the state
variables is used.
If a system is uncontrolable, no
state controller can be designed.
Firstly introduced by R.E. Kalman. R.E. Kalman
(1930-2016)
Adama Science and Technology
University

R.E. Kalman Decomposition of System

 Controllable and
observable system
 Controllable but
unobservable system
 Observable but
uncontrollable system
 Uncontrollable and
unobservable system
 5.3 Controllability
(Definition) Controllability
A system is said to be controllable at t= t0 if there is a
control input u(t), t0<tt1 that can move the initial
state x(t0) to the desired state x(t1) at t= t1.
5.3 Controllability
Single input LTI system:
The controllability is closely related to matrices A and
B. For example

0 1  
x  x    u, x (t0 )  0
0 1  

Transition matrix (t)

1 -1+et 
Φ (t )  e At  
0 e 
t
5.3 Controllability
The solution is 0
t

A ( t  t0 ) A(t  )
x (t )  e x (t0 )  e Bu ( )d
0

-1+e(t  )   
t 1
 0 0
 e (t  )



 
   u ( )d

 t 

(t  )
 0 [     e ]u ( )d 
 
 t


(t  )
e u ( )d
 0 
 5.3 Controllability
Letting t= t1
t1 t1
x1 (t1 )  (   )  0
u ( )d    0
e(t1  ) u ( )d
t1

(t1  )
x2 (t1 )   e u ( )d
0

When  0, = 0
t1
x1 (t1 )    0
u ( )d
x2 (t1 )  0
 x1(t1) can be moved to any state via u(t), but x2(t1) is
always fixed to 0.
 5.3 Controllability
(Definition) Controllability of a LTI system
A LTI system is said to be controllable at t= t0 if u(t),
t0< t  t1 exists to move x(t0) to the desired state x(t1)
at t= t1. Especially if one of the following conditions
is true, then it is controllable:

1. The rank of the nnm controllability matrix is n.

U= [B AB  An-1B]
2. All rows of the transfer matrix are linearly
independent.
(sI-A)-1B
5.3 Controllability
3. The controllable Gramian matrix WC is
nonsingular.
t

A T AT 
Wc  e BB e d , t  0
0
For example, given a system
0 1 1 0 
x  x  u, x (t0 )  x0
 1 2  0 1 
Controllability matrix is given by
1 0 0 1  Since U= 2 = n,
U  [ B AB ]   
 0 1 1 2  controllable.
 5.3 Controllability
Consider a system
 1 1 0  1 
   
x   0 1 0  x   2  u, x (t0 )  x0
 0 0 0  1 

Controllability matrix is given by

1 1 3
U  [ B AB A2 B ]   2 2 2 
1 0 0 
5.3 Controllability
1 1 3 -2E1+E2  E2
 
U  [ B AB A B ]   2 2 2 
2 -E1+E3  E3
1 0 0 

1 1 3 1
0 4 8  -4E2+E3  E3 1 1 3
  0 4 8 
0 1 3   
0 0 1 
Since U= 3= n, controllable.
 5.3 Controllability
Consider a LTI system
1 1  1
x  x    u, x (t0 )  x0
0 2  1
1
1  s  1 1  1
( sI  A) B    1
 0 s  2  
 s3 
1  s  2 1  1  ( s  1)( s  2) 
  0    
( s  1)( s  2)  s  1 1  1 
 ( s  2) 
 
To check the linear independence of two rows
s3 1
1  2 0
( s  1)( s  2) ( s  2)
 5.3 Controllability
Multiplying both sides by (s-1)(s+2) gives

1(s+3)+2(s-1)= (1+2)s+(31-2)= 0s+0s0

Comparing the left and right sides gives
1+2= 0 
31-2= 0 

Thus, 1= 2= 0. Since two rows of (sI-A)-1B are linearly

independent, controllable.
 5.3 Controllability
For example, given a system
 1 0  1
x  x    u, x (t0 )  x0
 0 2  1

Transition matrix (t)

 e t 0 
Φ (t )  e At  
2t
0 e 
t

A T AT 
Wc  e BB e d , t  0
0
5.3 Controllability
       2 3 
e 0 1  e 0 e e
e A BBT e A     1 1 
T
     
2      
0 e  1 0 e  e
2 3
e 
4

t t  e 2 e 3 


Wc  e A BBT e A  d   
T

   
d
0 e 3
e 4

0

1 2t 1 3t 
2 (1  e ) (1  e )
3
 
 1 (1  e3t ) 1 (1  e4t ) 
 3 4 
Since Wc is
1 1
Wc  (1  e2t )(1  e 4t )  (1  e 3t ) 2  0 nonsingular,
8 9
controllable.
 5.3 Controllability
For example, check the controllability using the
Matlab functions.
 1 1 0  1 
x   0 1 0  x   2  u, x (t0 )  x0
 0 0 0  1 
Program:
>>A= [-1 1 0;0 -1 0;0 0 0];
>> B= [1;2;1];
>> U= ctrb(A,B); % Controllability matrix
>> r= rank(U) % Rank
r= 3
5.4 Observability
Why do we need to check it?
Designing and implementing state feedback control
require the feedback of all state variables.
When some state variables are unmeasurable, we
need to check whether they can be estimated from the
model.
If the system is unobservable, state observer design is
impossible
Firstly introduced by R.E. Kalman
5.4 Observability
LTI system:
The observability is closely related to matrices A
and C.
x  Ax  Bu, x (t0 )  x0
y  Cx  Du

x  Ax  Bu, x (t0 )  x0
y  Cx
 5.4 Observability
(Definition) Observability
A LTI system is said to be observable if there is
enough t= t1>0 that u and y determine x0 uniquely.
Especially it is observable if one of the following
conditions is true:
1. The rank of the npm observability matrix V is n.

 C 
 CA 
V  
 
 n-1 
CA 
 5.4 Observability
2. All columns of the transfer matrix are linearly
independent.

C(sI-A)-1

3. The observable Gramian matrix WO is nonsingular.

t

AT 
Wo  e C T Ce A d , t  0
0
 5.4 Observability
For example, consider a SISO LTI system
0 1 0
x  x    u, x (t0 )  x0
 1 2  1 
y  1 0 x
Observability matrix yields
 C  1 0 
V   
CA 0 1 
 Since V= 2 = n, the system is observable.
5.4 Observability
For example, consider a LTI system

0 1  1 
x  x    u, x (t0 )  x0
0 0  0
y  1 0 x
1
 s 1
C ( sI  A)  1 0 
1

 0 s 
 s 1  1 1
 2 1 0 
1
  2
s 0 s   s s 
 5.4 Observability
To check the linear independence of two columns
1 1
1   2 2  0
s s
Multiplying both sides by s2 gives

1s+2= 0s+0s0
Comparing the left and right sides gives

 Since 1= 2= 0, the system is observable.

5.4 Observability
Program:
>> A= [0 1;0 0]; B= [1;0]; C= [1 0];
>> U= ctrb(A,B); % Controllability matrix
>> ru= rank(U) % Rank
>> V= obsv(A,C); % Observability matrix
>> rv= rank(V) % Rank
ru= 1
0 1  1 
rv= 2 x  x    u, x (t0 )  x0
0 0  0
y  1 0 x
The system is observable but not controllable.
 5.5 Similarity Transformation
A LTI system can be represented as a combination of
subsystems with appropriate similarity transformation
Controllable subsystem
x  Ax  Bu, x (t0 )  x0
y  Cx  Du
If U= n1< n, via the following similarity
transformation
x  Qx
The original system can be represented by a
combination of controllable subsystems and
uncontrollable subsystems.
5.5 Similarity Transformation

(n-n1) vectors randomly chosen

such that Q is nonsingular

where Q= [q1 q2 … qn1 qn1+1 … qn]

n1 independent linear vectors

selected from U
 5.5 Similarity Transformation
Original system
x  Ax  Bu, x (t0 )  x0
y  Cx  Du
Similarity transformation x  Qx

x  Ax  Bu
y  Cx  Du
where A  Q 1 AQ, B  Q 1 B, C = CQ, D  D
5.5 Similarity Transformation
Transformed system
 xc   Ac A12   xc   Bc 
    u
 xc   0 Ac   xc   0 
 xc 
y  Cc C c     Du
 xc 
Controllable (stabilizable or reachable) subsystem

xc  Ac xc  Bc u
y  Cc xc  Du
5.5 Similarity Transformation
Consider a LTI system.

0 1 0 0 

x   1   
2 0  x  1  u, x (t0 )  x0
 0 0 3 0 
y  1 1 0 x
The controllability matrix U gives
0 1 2 
U  [ B AB A2 B ]  1 2 3 
0 0 0 
5.5 Similarity Transformation
Since U= 2 < n= 3, it is uncontrollable. n1= 3-2= 1.
Choosing one vector randomly such that Q is
nonsingular yields

0 1 2  0 1 0 
 
U  1 2 3  Q  1 2 0 
0 0 0  0 0 1 
2 1 0
Q 1 
 1 0 0  
Randomly choose
 0 0 1  such that Q-1 exists
 5.5 Similarity Transformation
Thus
0 - 1 0
A  Q 1 AQ  1 -2 0  ,
0 0 -3 
1 
1 
B  Q B  0  , 
0 
C  CQ  1 1 0 ,
DD0
 5.5 Similarity Transformation
The transformed system is given by
0 - 1 0  1 
 xc     xc   
   1 -2 0     0  u
 xc    xc 
0 0 -3  0 
 xc 
y  1 1 0  
 xc 
0 1 1 
Thus, controllable xc    xc    u
subsystem is given by 1 2  0
y  1 1 xc
 5.5 Similarity Transformation
Observable subsystem
x  Ax  Bu, x (t0 )  x0
y  Cx  Du
If V= n2 < n, via the similarity transformation
x  Q 1 x
the original system can be represented by a
combination of observable subsystems and
unobservable subsystems.
 5.5 Similarity Transformation

 q1 
 q  n2 independent vectors
 2 
  selected from V
 
where Q   qn2 
q 
 n2 1  (n-n2) vectors randomly
  chosen such that Q is
  nonsingular
 qn 
 5.5 Similarity Transformation
Original system
x  Ax  Bu, x (t0 )  x0
y  Cx  Du
Similarity transformation x  Q 1 x
x  Ax  Bu
y  Cx  Du
where A  QAQ 1 , B  QB, C  CQ 1 , D  D
5.5 Similarity Transformation
Transformed system
 xo   Ao A12   xo   Bo 
    u
 xo   0 Ao   xo   0 
 xo 
y  Co Co     Du
 xo 
Observable subsystem
xo  Ao xo  Bo u
y  Co xo  Du
 5.5 Similarity Transformation
Consider a LTI system.
0 1 0 0 

x   1   
2 0  x  1  u, x (t0 )  x0
 0 0 3 0 
y  1 1 0 x
The observability matrix V gives
 C   1 1 0
   
V   CA    1 1 0 
CA2   1 1 0 
   
 5.5 Similarity Transformation
Since V= 1 < n= 3, it is unobservable. Selecting Q as
1 1 0  1 1 0 
Q  0 1 0  Q 1  0 1 0 
0 0 1  0 0 1 
Randomly choose
Thus such that Q-1 exists
-1 0 0  1 
A  QAQ 1  -1 -1 0  , B  QB  1 
0 0 -3  0 
C  CQ 1  1 0 0 , D  D  0
 5.5 Similarity Transformation
The transformed system is given by
-1 0 0  1 
 xo     xo   
   -1 -1 0    xc  1  u
 xo    xo 
0 0 -3  0 
 xo 
y  1 0 0  
 xo 
Thus, observable ?  1 0  1
xo    xo    u
subsystem is given by  1 1 1
y  1 0 xo
 5.6 Controllable Canonical Form
State feedback control
d
r Feedforward + u + + y
Plant
gain matrix -
Feedback x
gain matrix
 5.6 Controllable Canonical Form

Plant
Digital Controller u y
r
u   Kx  Fr D/A

A/D A/D
x1 x2
5.6 Controllable Canonical Form
Plant
x  Ax  Bu, x (t0 )  x0
y  Cx  Du
where A  nn , B  n1 , C  1n ,D  
Assumption
The plant is a SISO LTI system(u and y are scalars)
It is controllable & observable(U= n, V= n)
5.6 Controllable Canonical Form
Similarity transformation
Before
x  Ax  Bu, x (t0 )  x0
y  Cx  Du
x  Qx
After
Q is
x  Ax  Bu , x (t0 )  x0 nonsingular
y  Cx  Du
1 1
where A  Q AQ, B  Q B, C  CQ , D =D
 5.6 Controllable Canonical Form
Controllability
Before U  [ B AB An 1 B ], U  n
After U  [ B AB Ai 1 B An 1 B ]
The ith term in the above equation

Ai 1 B  (Q 1 AQ )(Q 1 AQ ) (Q 1 AQ )Q 1 B  Q 1 Ai 1B

Thus (i-1) times

U  Q 1[ B AB Ai 1 B An 1 B ]  Q 1U , U  n
If Q is nonsingular, controllability is kept after transformation.
 5.6 Controllable Canonical Form
Observability
 C   C 
 CA   
V   , V  n After 
V 
CA 
Before   
 n 1   
CA  CA 
n1

As the previous case

CAi 1  CQ (Q 1 AQ )(Q 1 AQ ) (Q 1 AQ )  CAi 1Q

(i-1) times
 5.6 Controllable Canonical Form
After transformation
 C   C 
   
   
V   CAi 1    CAi 1  Q  VQ
   
   
 n 1   n 1 
CA  CA 
Thus V  n
 Observability is kept after transformation.
 5.6 Controllable Canonical Form
Conversion to controllable canonical form
System
x  Ax  Bu, x (t0 )  x0
y  Cx  Du
Controllable canonical form
x  Ax  Bu, x (t0 )  x0
y  Cx  Du
where A  Q 1 AQ, B  Q 1 B, C  CQ , D =D
 5.6 Controllable Canonical Form
Controllable canonical form(SISO)
 0 1 0 0  0 
 0 0 1 0  0 
   
x  x    u, x (t0 )  x0
   
 0 0 0 1  0 
 a a a1  1 
 n n 1  an  2

y  (bn  b0 an ) (bn 1  b0 an 1 ) (b1  b0 a1 )  x  b0 u

Transfer function

(b1  b0 a1 ) s n 1  (b2  b0 a2 ) s n 1   (bn  b0 an )

G(s)  n 1
 b0
s  a1s
n
  an 1s  an
 5.6 Controllable Canonical Form
Conversion procedure
Step 1: Check the controllability of the system (U= n).
U  [ B AB An1 B ]
Step 2: Find the characteristic polynomial from A.
( ) |  I  A |  n  a1 n 1   an 1  an
Step 3: Obtain Q= [q1 q2 … qn]
qn  B
qn 1  AB  a1 B
qn  2  A2 B  a1 AB  a2 B

q1  An 1 B  a1 An  2 B  an 1 B
5.6 Controllable Canonical Form
Step 4: Obtain matrices
1 1
A  Q AQ, B  Q B, C  CQ , D =D
For example, consider a dynamic system.
1 0 1
x  x u
 1 2  1
y   2 1 x
Check the controllability.
Convertible to the
1 1 
U  [ B AB ]    , U  2 controllable
1 3 canonical form.
5.6 Controllable Canonical Form
Getting the characteristic polynomial of A
 1 0
 ( )   I  A     2
2
1 2
a1= 1, a2= -2 1 0 1
x  x u
 1 2  1
1 y   2 1 x
q2  B   
1
1 1  2 
q1  AB  a1 B     1      Q= [q1 q2]
 3 1  2 
 5.6 Controllable Canonical Form
 2 1 1 1 1 1
Q   Q   
 2 1 4 2 2 
Obtain matrices

1 1 1 1  1 0   2 1  0 1 
A  Q AQ =        
4  2 2   1 2   2 1  2 1
1 1 1 1 1 0 
B  Q B=      
4  2 2  1 1 
 2 1
C  CQ =  2 1     2 3 , D  D =0
 2 1
5.6 Controllable Canonical Form
Before
1 0 1 G ( s )  C ( sI  A) 1 B  D
x  x   u
 1 2  1 3s  2
 2
y   2 1 x s s2
After
0 1  0
x  x   u G ( s )  C ( sI  A) 1 B  D
 2 1 1 
3s  2
y   2 3 x  2
s s2
 TFs before and after the
transformation are the same.
 5.6 Controllable Canonical Form
Example 1: Code to obtain the controllable canonical
form of the system.
 2 1 0  4
x   0 0 1  x   0  u
 1 0 0  1 
y  1 0 0 x

Program:
 5.6 Controllable Canonical Form
MATLAB program
A= [-2 1 0;0 0 1;-1 0 0]; B= [4;0;1]; C= [1 0 0];
D= 0;
p= poly(A); %  ( )    a1  a 2   a 3
3 2

Q(:,3)= B;
Q(:,2)= A*Q(:,3)+p(2)*B;
Q(:,1)= A*Q(:,2)+p(3)*B;
IQ= inv(Q);
A1= IQ*A*Q, B1= IQ*B, C1= C*Q, D1= D
[ns,ds]= ss2tf(A,B,C,D) % TF of original
[ns1,ds1]= ss2tf(A1,B1,C1,D1) % TF after
 5.6 Controllable Canonical Form
Example 2: Obtain the TF of the system.
0 1 0 0 
x   0 0 1  x  0  u
 2 0 1 1 
y   2 3 0 x
Solution:
G ( s )  C ( sI  A) 1 B  D
b1s 2  b2 s  b3 3s  2
 
s 3  a1s 2  a2 s  a3 s3  s 2  2
 5.6 Controllable Canonical Form
Example 3: Write the TF in the controllable
canonical form.
2 s  3s  6s  2
3 2
G (s) 
s 3  6 s 2  5s  1
Solution:
a1= 6, a2= 5, a3= 1, b0=2, b1= 3, b2= 6, b3= 2
 0 1 0  0  0 1 0 0 
x   0 0 1  x  0  u   0 0 1  x  0  u
 a3 a2 a1  1   1 5 6  1 
y  (b3  b0 a3 ) (b2  b0 a2 ) (b1  b0 a1 )  x   0 4 9 x
 5.7 Observable Canonical Form
Repeat Example 3 in the Observable canonical form

0 0 0 an   (bn  b0 an ) 
1   
 0 0 an 1   (bn 1  b0 an 1 ) 
x  0 1 0 an  2  x  (bn  2  b0 an  2 )  u, x (t0 )  x0
   
   
0 1 a1   (b  b a ) 
 0  1 0 1 
y  0 0 0 1 x  b0u
Adama Science and Technology
University

Assignment
From Linear System Theory and Design: 6.1
, 6.10, 6.14, 6.15, 6.16, 6.20, and 6.21
Note: validate your solution using matlab
Q&A