EI6801- Computer control of Process

UNIT I DISCRETE STATE-VARIABLE TECHNIQUE 9

State equation of discrete data system with sample and hold – State transition
equation – Methods of computing the state transition matrix – Decomposition of
discrete data transfer functions – State diagrams of discrete data systems – System
with zero-order hold – Controllability and observability of linear time invariant
discrete data system–Stability tests of discrete-data system – State Observer - State
Feedback Control.
State Space Representation of Discrete time system:
The general form of state model for an nth order linear time invariant discrete time
system is,
X  k  1  F X  K   g u  k   state equation
Y k   C X k   D u k   output equation
where, X  k   n 1 state vector
y  k   p 1 output vector
u  k   m 1 input vector
g  nm input matrix
C  pn output matrix
D  pm transmission matrix; F  n  n system matrix

Conversion of state variable models to transfer function:

X  k  1  F X  k   g u  k 
Y k   C X k   D u k 

Taking Z transform on both sides,

ZX  z  -ZX  0  =F X  z  +g U  z 
Y  z  =CX  z  +D U  z 

X  z  =  ZI - F Z X  0  +  ZI-F g U  z 
-1 -1

Now Y  z  =C  ZI-F  Z X  0  + C  ZI-F  g+D  U  z 

-1 -1
 
with zero initial condition, (ie) X(0) = 0,
Y z adj  ZI  F 
=C  ZI  F  g+D Where  ZI  F  
1 1

U z  ZI  F 
Solution of homogeneous equation:
X (k  1)  F X (k )  (1) is the homogeneous state equation.

The solution is, X (k )  F k X (0)  (2)

Taking Z transform on the both sides of equation 1,

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.1

 EI6801- Computer control of Process
Z X ( z )  Z X (0)  FX ( Z )
X ( Z )  ( ZI  F ) 1 Z X (0)
X (k )   1[( ZI  F ) 1 Z ] X (0)

Comparing with equation (2)

F k  1[(ZI  F )1 Z ]   (k ) ------ This is the state transition matrix

Solution of Non homogeneous equation:
X  k  1  F X  k   g u  k  is the non homogeneous state equation.

At k=0,1 we get
X 1  F X  0   g u  0 

X  2   F X 1  g u 1

Substituting X 1 in X  2

X  2  F  F X  0   g u  0   g u 1  F 2 X  0   Fg u  0   g u 1  (1)

Rewriting equation (1) we get,

2 1
X  2   F 2 X  0    F 2i 1 g u  i 
i 0

In general,
k 1
X  k   F k X  0    F k i 1 g u  i   (2)
i 0

The matrix F k is the state transition matrix.

The solution given in equation (2) contains two terms. The first term is the response
due to zero input and the second term is the response due to zero initial conditions.
Conversion of transfer functions to canonical state variable models:
The general transfer function is,
Y  z  b0 z n +b1 z n 1  ......  bn 1 z  bn
= n
u z z +a1 z n 1  ......  an 1 z  an

Controllable phase variable form:

The matrix ‘F’ has a very special structure the co-efficient of the denominator
of the transfer function preceded by minus sign form a string along the bottom row
of the matrix. The rest of the matrix is zero except for the super diagonal terms which
are all unity.

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.2

 EI6801- Computer control of Process
0 1 0 ......... 0  0 
  0 
0 0 1 0   
F  .  g  :  C  bn bn1 ........ b0  D  b0
   
0 0 0 1  0 
 a  an 1  an  2 
 a1  1 
 n
If the degree of the numerator is equal to the degree of the denominator then,

C  bn  anbo bn1  an1b0 ..... b1  a1b0 

Observable phase variable form:

The coefficient of the denominator of the transfer function appears in one of
the column of the F matrix.
0 0  0  an 
   bn 
1 0  0  a n 1  b 
g C   0 0  1
n 1 
F  0 1 0  an 2 
   
      
0 0 1  a 1   b0 


Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.3

 EI6801- Computer control of Process

Diagonal canonical form:

The partial fraction expansion of the transfer function is,

Y  z r1 r2 rn
 b0    ... 
U  z Z  p1 Z  p2 Z  pn

The coefficients ri  i  1, 2,...n  are the residues of the transfer function at the

corresponding poles at z  pi  i  1, 2,...n  .

 p1 0 . . 0 1
0 p2 . . 0  1
 
F . . . . .  ; g   .  ; C   r1 r2 . . rn 
  
. . . . .  .
 0 0 . . pn  1

r1 r2 rn
Y  z   b0U  z   U  z  U  z   ...  U  z
Z  p1 Z  p2 Z  pn
U (Z )
X1 ( k )   X ( k  1)  p1 X1 ( k )  U ( k )
Z  p1 1

U (Z )
X 2 (k )   X ( k  1)  p2 X 2 ( k )  U ( k )
Z  p2 2

The block schematic is shown in Fig below

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.4

 EI6801- Computer control of Process

Jordan canonical form: (if there are repeated poles)

It is assumed that the transfer function G  z  has m distinct poles at z = p1,p2.----pm of

multiplicity n1,n2.----nm respectively. Then, G  z  is of the form,

b1'zn-1 + b2'zn-2 + ..... + bn'

G  z  = b0 +
 z - p1   z - p2  .....  z - pm 
n1 n2 nm

The partial fraction expansion of G  z  is of the form,

G z  = b0 +H1(z)+......+Hm (z)

Where,
ri1 ri2 rin
Hi (z) = + +... +
 z - pi 
ni
 z - pi 
ni -1
 z - pi 
The first term in H i ( z ) can be synthesized as a chain of ni identical first order terms

1
each having transfer function . The second term can be synthesized as a chain
 Z  pi 
of ni  1 first order terms. The entire H i ( z ) can be synthesized by the system having

the block diagram shown in Fig below.

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.5

 EI6801- Computer control of Process
-p1 1 0 . 0 . . 0  0 
0 -p1 1 . . . . .   0 
  
0 0 -p1 . 0 . . 0   1
   
. . . . . . . .  ...
F= ; g =   ; C = r1 r2 . . rn 
0 0 0 . -p 4 . . 0   1
   
 . . . . . . . .  .
 . . . . . . . .  .
   
 0 0 0 . 0 . . -pn   1 

Y  z  4z3 -12z 2 +13z-7

Example 1.1: Consider =
u z ( z  1) 2 ( z  2)

Obtain 3 different state models and Draw the block diagram for the state model.
Solution:
Y  z  4z3 -12z 2 +13z-7
=
u  z  z3  4 z 2  5z  2

Controllable phase variable form:

 X 1 (k  1)  0 1 0   X 1 ( K )  0 
 X (k  1)   0  
 2   0 1   X 2  K    0  U ( K )
 X 3 (k  1)   2  5 4   X 3  K   1 
 

 X1 ( K ) 
 
Y ( K )  1  7 4   X 2  K    4U ( K )
X K 
 3  
Elements of C matrix:
b0  4; b1  12;
b2  13; b3  7; a1  4; a2  5; a3  2

b3  a3b0  7  (2)*4  7  8  1

b2  a2b0  7 & b1  a1b0  4

X ( k  1)
1
 X 2 (k )

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.6

 EI6801- Computer control of Process
X ( k  1)
2
 X 3 (k )

X ( k  1)
3
 2 X1 (k )  5 X 2 ( k )  4 X 3 ( k )  U ( k )

Y (k )  X1 (k )  7 X 2 (k )  4 X 3 (k )  4U (k )

Observable phase variable form:

 X 1 (k  1)  0 0 2   X 1 ( K )  1 
 X (k  1)   1  
 2   0  5   X 2  K     7  U ( K )
 X 3 (k  1)  0 1 4   X 3  K    4 
 

 X1 ( K ) 
 
Y ( K )  0 0 1   X 2  K    4U ( K )
X K 
 3  
X ( k  1)
1
 2 X 3 (k )  U (k ) X ( k  1)
2
 X1 (k )  5 X 3 (k )  7U (k )

X ( k  1)
3
 X 2 ( k )  4 X 3 (k )  4U (k ) Y (k )  X 3 ( k )  4U ( k )

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.7

 EI6801- Computer control of Process
Jordan canonical form :
Y ( z) 4z2  7z 1
 4
U ( z) ( z  1) 2 ( z  2)
4z2  7z 1
Applying p. f .e. to
( z  1) 2 ( z  2)
Y ( z) 2 1 3
Now,  4  
U ( z) ( z  1) 2
( z  1) ( z  2)

 X 1 (k  1)  1 1 0   X 1 ( K )  0
 X (k  1)   0    
 2   1 0   X 2  K    1  U ( K )
 X 3 (k  1)  0  X K  1 
 3    
0 2 

 X 1 (k ) 
Y ( K )  [2 1 3]  X 2 (k )   4 U ( K )
 X 3 (k ) 

Signal flow graph:

X 1 ( K  1)  X 1 ( K )  X 2 ( K )
X 2 ( K  1)  X 2 ( K )  U ( K )
X 3 ( K  1)  2 X 3 ( K )  U ( K )

Y ( K )  2 X1 ( K )  X 2 ( K )  3 X 3 ( K )  4U ( K )

z+5
Example 1.2: Consider G(z) =
z  12 z 2  50 z  30
3

Obtain controllable and observable phase variable form.

Solution:
z+5
G(z) =
z  12 z 2  50 z  30
3

Controllable phase variable form:

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.8

 EI6801- Computer control of Process
 X 1 (k  1)  0 1 0   X 1 ( K )  0 
 X (k  1)   0     
 2   0 1   X 2  K    0  U ( K )
 X 3 (k  1)   30  50  X K  1 
 3    
 12 

 X1 ( K ) 
 
Y ( K )  5 1 0   X 2  K  
X K 
 3  
Observable phase variable form:

 X 1 (k  1)  0 0  30   X 1 ( K )  5 
 X (k  1)   1  
 2   0  50   X 2  K    1  U ( K )
 X 3 (k  1)  0 1  12   X 3  K   0
 

 X1 ( K ) 
 
Y ( K )  0 0 1   X 2  K    4U ( K )
X K 
 3  
Example 1.3: Obtain the Jordan canonical form realization for the following transfer
function.

Y ( z) z 3  8 z 2  17 z  8

U ( z )  z  1 z  2  z  3

Jordan canonical form : The given function is converted into a proper fraction.

Y ( z) 2z 2  6z  2
 1
U ( z) ( z  1)( z  2) ( z  3)

2z2  6z  2
Applying p. f .e. to
( z  1)( z  2) ( z  3)

2z2  6z  2 A B C
  
( z  1)( z  2) ( z  3) ( z  1) ( z  2) ( z  3)

Solving for constants, A=-1,B=2,C=1

Y ( z) 1 2 1
Now,  1  
U ( z) ( z  1) ( z  2) ( z  3)

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.9

 EI6801- Computer control of Process
 X 1 (k  1)   1 0 0   X 1 ( K )  1
 X (k  1)   0  2 0    
 2     X 2  K    1 U ( K )
 X 3 (k  1)  0 0  3    1
 X3  K   
 X 1 (k ) 
Y ( K )  [1 2 1]  X 2 (k )   U ( K )
 X 3 (k ) 

Signal flow graph:

X 1 ( K  1)   X 1 ( K )  U ( K )
X 2 ( K  1)  2 X 2 ( K )  U ( K )
X 3 ( K  1)  3 X 3 ( K )  U ( K )

Y ( K )  4 X1 ( K )  2 X 2 ( K )  X 3 ( K )  U ( K )

The SFG of the state model is,

Example 1.4: For the difference equation, obtain Jordan canonical form realization
y( K  3)  5 y( K  2)  7 y( K  1)  3 y( K )  r ( K  1)  2r ( K )

Solution:

Taking Z Transform on both sides,

Y ( z )  Z 3  5Z 2  7Z  3  R(Z )  Z  2

z+2 z+2
G(z) = 
z 3  5 z 2  7 z  3  z  12  z  3

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.10

 EI6801- Computer control of Process
z+2
Applying partial fraction expansion to
 z  1  z  3
2

z+2 A B C
Now,   
 z  1  z  3 ( z  1) ( z  1) ( z  3)
2 2

Solving for constants A=0.25, B=0.5,C=-0.25

 X 1 (k  1)   1 1 0   X 1 ( K )  0   X 1 (k ) 
 X (k  1)   0  1    
 2   0 
  X 2  K    1  U ( K ) Y ( K )  [0.25 0.5  0.25]  X 2 (k ) 
 X 3 (k  1)  0 0  X K  1   X 3 (k ) 
 3    
3 

Signal flow graph:

X 1 ( K  1)   X 1 ( K )  X 2 ( K )
X 2 ( K  1)   X 2 ( K )  U ( K )
X 3 ( K  1)  3 X 3 ( K )  U ( K )

Y ( K )  0.25 X1 ( K )  0.5 X 2 ( K )  0.25 X 3 ( K )

Methods to determine state transition matrix:

i) Similarity transformation method
ii) Cayley Hamilton method
iii) Inverse Z transform method
Example 1.5: Consider the system

 X 1 (k  1)  0 1   X1 ( K )  with X 1 (0)  X 2 (0)  1

 X (k  1)    2  3   
 2     X 2  K  Y (k )  X 1 (k )

Find state transition matrix (k )

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.11

 EI6801- Computer control of Process
Solution:
0 1 
Given F   
 2  3 
Inverse Z transform method:

 (k )  1  ZI  F  Z 
1
 

Z 0   0 1  Z 1 
 ZI  F     
0 Z   2 3  2 Z  3

1 Z  3 1 
 ZI  F 
1

z ( z  3)  2  2 Z 

 ( Z  3) Z Z 
 Z 2  3Z  2 Z  3Z  2 
2
Now,  (k )   1  
 2Z Z2 
 Z 2  3Z  2 Z 2  3Z  2 

X1 ( z)  Z 3   Z 3 
  1  2    1  
z  Z  3Z  2    Z  1 Z  2  

x1 (k )    2   2(1)k
k

X 2 ( z)  1   1 
  1  2    1  
z  Z  3Z  2    Z  1 Z  2  

x2 (k )    2   (1)k
k

X 3 ( z)  2   2 
  1  2   1
 
z  Z  3Z  2    Z  1 Z  2  

x3 (k )  2  2   2(1)k
k

X 4 ( z)  Z   Z 
  1  2   1
 
z  Z  3Z  2    Z  1 Z  2  

x4 (k )  2  2   (1)k
k

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.12

 EI6801- Computer control of Process
   2 k  2(1) k   2   (1) k 
k

 (k )   
 2  2   2(1)k 2  2   (1) k 
k k

Similarity transformation method

(k )  Pk P 1

P is the transformation matrix that transforms matrix A into diagonal form.

1 1 
P Where the Eigen values are 1  1, 2  2
1 2 

 2 1  0 1   1 1   1 0 
The matrix   P 1FP  (1)    2 3  1 2   0 2
 1 1     
(1) k 0 
Now k=  
 0 (2) k 

 1 1  (1)k 0   2 1   2(1) k  (2) k (1) k  (2) k 

(k )      
 1 2  0 (2)k   1 1  2(1)k  2(2) k (1) k  2(2) k 

Computation of the state transition matrix using Cayley Hamilton Theorem:

Step 1: Determine Eigen values of given matrix ‘F’
Step2:Let the matrix polynomial be,

g     0 I  1  .....  n1 n1

The coefficients 0, 1, 2 .... n1 are evaluated using the following equation

f  i   i k  g  i 

Case i) Two distinct Eigen values: 1  2

g     0 I  1

The coefficients 0 and 1 are evaluated using the following equations

f  1   1k  g  1 

f  2   2k  g  2 

Case ii) Two equal and one unequal Eigen values: 1  2 , 3

g     0 I  1  2 2

The coefficients 0 , 1 and 2 are evaluated using the following equations

f  1   1k  g  1 
i)

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.13

 EI6801- Computer control of Process
 d   d 
ii)  f     g   
 d   2  d    2

iii) f  3   3k  g  3 

The state transition matrix   k   0 I  1F  2 F 2  ....

Example 1.6:
0 1
Given F    Find state transition matrix (k ) using Cayley Hamilton
 1  2 
theorem.
Solution:

 0  0 1   1 
The Eigen values are,  I  F          2  2  1  0
 0    1  2   1   2
  1, 1

Since ‘F’ is second order, g     0 I  1

The coefficients 0 and 1 are evaluated using the following equations

f  1   1  0  1
k

 d   d 
 d  f       d  g    
 1  1

 d k  d 
 d   1    d   0  1 

 k  1k 1   1  1  k  1
k
 

0   1  1   1  k  1   1 1  k 

k k k k

The state transition matrix   k   0 I  1F

 1k 1  k  0  k 0 1
 k      k  1  
  1 1  k   1  2 
k
0

1  k k 
  k    1 
k

k 1  k 

Example 1.7: A system is described by the state equation

 2 1 0   3 
X (k  1)   4 0 1  X (k )   7  U (k ); X (0)  x0

 1 0 0   0 

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.14

 EI6801- Computer control of Process
Using Z transform technique, transform the state equation into a set of linear
algebraic equations in the form X ( z)  G( z) X (0)  H ( z)U ( z) . Also draw a block

diagram realization of the state equation.

G( z )   ZI  F  Z 
1
 

 z  3 1 0 
 ZI  F    4 z 1
 1 0 z 

 ZI  F   z 3  3z 2  4 z  1  

 z2 z 1 
1 
 ZI  F  Z    4 z  1 z  z  3
1
z  3  Z  G( z)

 z
 1 z  z  3  4 

 z2 z 1   2 
1 
H ( z )   ZI  F  g     4 z  1 z  z  3
  z  3  7 
1
  
 z
 1 z  z  3  4   0 

 3z 2  7 z 
1 
H ( z )   7 z 2  9 z  3

 3z  7 
 
Example 1.8: Given

 0 1
F  
 4 5
Determine the state transition matrix using
i) the Z transform method ii) Similarity transformation iii) Cayley Hamilton
method
Solution:
 0 1
Given F   
 4 5
Inverse Z transform method:

 (k )  1  ZI  F  Z 
1
 

Z 0   0 1  Z 1 
 ZI  F     
0 Z   4 5  4 Z  5

1 Z  5 1 
 ZI  F 
1

z ( z  5)  4  4 Z 

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.15

 EI6801- Computer control of Process
 ( Z  5) Z Z 
 Z 2  5Z  4 Z  5Z  4 
2
Now,  (k )   1  
 4Z Z2 
 Z 2  5Z  4 Z  5Z  4 
2

X1 ( z)  ( Z  5)   Z 5 
  1  2    1  
z  Z  5Z  4    Z  1 Z  4  

x1 (k )  1.331  0.33(4)k
k

X 2 ( z)  1   1 
  1  2   1
 
z  Z  5Z  4    Z  4  Z  1 

x2 (k )  0.331  0.33(4)k
k

X 3 ( z)  4   4 
  1  2   1
 
z  Z  5Z  4    Z  1 Z  4  

x3 (k )  1.32 1  1.32(4)k

k

X 4 ( z)  Z   Z 
  1  2    1  
z  Z  5Z  4    Z  1 Z  4  

x4 (k )  0.331  1.33(4)k
k

1.33 1k  0.33(4)k 0.33 1  0.33(4)k 

k

 (k )   
1.32 1  1.32(4)k 0.33 1  1.33(4)k 
k k

Similarity transformation method

(k )  Pk P 1
P is the transformation matrix that transforms matrix A into diagonal form.

1 1 
P Where the Eigen values are 1  1, 2  4
1 2 

 4 1  0 1 1 1  1 0 
The matrix   P 1 AP  (0.33)     
 1 1   4 5 1 4 0 4
(1)k 0 
Now k=  
 0 (4)k 

0   1.32 0.33 1.331  0.33(4) 0.331  0.33(4) k 

k k
1 1  (1)k
k

(k )      
1 4  0 (4)k   0.33 0.33  1.32 1k  1.32(4)k 0.331  1.33(4)k 
k

Cayley Hamilton theorem

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.16

 EI6801- Computer control of Process
 0  0 1   1 
The Eigen values are,  I  F         2  5  4  0
 0    4 5   4   5

Eigen values are 1  1, 2  4

Since ‘F’ is second order, g     0 I  1

The coefficients 0 and 1 are evaluated using the following equations

f 1  1  0  1
k

f  4    4   0  4 1
k

31   4   1  1  0.33  4   0.33 1

k k k k

0  1  1  1  0.33  4   0.33 1   0.33  4   1.33 1

k k k k k k
 

The state transition matrix   k   0 I  1F

 0.33  4 k  1.33 1k 0  0 1

 k      0.33  4   0.33 1
k k
 4 5 
 0.33  4   1.33 1 
k k
0 

1.33 1k  0.33(4)k 0.33 1  0.33(4)k 

k

 (k )   
1.32 1  1.32(4)k 0.33 1  1.33(4) 
k k k

Diagonalization
Example 1.9: Consider a discrete time system with

0 1 0 0 

X (k  1)   0 0 1  X (k )  0  U (k )

 6 11 6  1 
y   6 0 0 X (k )

Obtain the state model in diagonal canonical form.

Solution:
The characteristic equation is

 1 0
I  F  0  1  0
6 11   6

(  1)(  2)(  3)  0

The Eigen values of matrix ‘F’ are 1  1, 2  2, 3  3

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.17

 EI6801- Computer control of Process
Eigen vectors V1 ,V2 ,V3 corresponding to the Eigen values 1 , 2 & 3 respectively can

be determined from the ad joint matrix i I  F

 (  6)  14  6 2
Now, adj  I  F  3(  6)  24  (  6) 2
21  12 (7  12) ( 2  3)

9 5 2 1

For 1  1, adj  1 I  F    9 5 2 
  
Eigen vector V1   1
 9 5 2   1

 6 4 2 2
   
For 2  2, adj[2 I  F ]   12 8 4  Eigen vector V2   4 
 3 2 1   1 

 5 3 2 1
   
For 3  3, adj[3 I  F ]   15 9 6  Eigen vector V3   3
 15 9 6   3 

The model matrix M is obtained by placing the Eigen vectors together as given by

1 2 1
M   1 4 3
 1 1 3 

 1 0 0 
F  M 1 FM   0 2 0 
 0 0 3

Controllability:
For the linear system given by equation,
X (k  1)  F X (k )  gU (k )
y ( k )  c X (k )  d U (k )

If there exists an input u(k), which transfers the initial state X(0) = X 0 to the state X1 at
k=N, the state X0 is said to be controllable. If all initial states are controllable, the
system is said to be completely controllable. Otherwise, the system is said to be
uncontrollable. The necessary and sufficient condition for the system to be
completely controllable is that the nxn controllability matrix, Q c = [g Fg F2g…. Fn-1g]

has rank equal to n. i.e., rank( Q c ) = n.

Observability:
For the linear system given by equation,
Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.18
 EI6801- Computer control of Process
X (k  1)  F X (k )  gU (k )
y ( k )  c X (k )  d U (k )

If it is possible to determine the state X(0) = X0 by the measurement of output

Y(k) over a finite time interval then the state X0 is said to be observable. If all initial
conditions are observable, the system is said to be completely observable. Otherwise,
the system is said to be unobservable. The necessary and sufficient condition for the
system to be completely observable is that the nxn observability matrix,

Qo  C CF CF2 .. ... CFn 1  T has rank equal to n, ie) rank ( Q o ) = n.

Example 1.10: Examine the controllability and observability property of the

following system

 1 1  0
X (k  1)    X (k )    U (k )
 0 1 1 
y (k )  1 1 X (k )

Solution:
The controllability matrix is, Qc   g Fg 

 1 1  0  1 
Fg       
 0 1 1   1
0 1 
 Qc    & Qc  1
1 1
The system is therefore completely controllable.

The observability matrix is Q0  C T F T C T 

 1 0  1  1
F T CT       
 1 1 1  0 
1 1
 Q0    & Qc  1
1 0 
The system is therefore completely observable.
Example 1.11: Investigate the controllability and observability of the following
system,
1 2 1 1 1 0 
X (k  1)    X (k )    U (k ); y (k )    X (k )
1 1 0 0  0 1 
Controllable:

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.19

 EI6801- Computer control of Process
The necessary and sufficient condition for the system to be completely controllable is

that the n*n controllability matrix, Qc   g Fg  has rank = n.

1 2 1 1
F   ;g  
1 1 0 0 
1 1 1 2  1 1 1 1
g   ; Fg     
0 0  1 1 0 0  1 1
1 1 1 1
Qc   
0 0 1 1
1 1
The determinant of one of the sub matrix is = -1. & rank( Qc ) =2. So the given
0 1

system is completely controllable.

Observable:
C
Q0   
CF
1 0  1 2
c   ; F  1 1
0 1   
1 0 
0 1 
1 0  1 2  1 2   
cF    1 1  1 1 ;V  1 2 
 0 1    
 
1 1
1 0
The determinant of one of the sub matrix is = 1. & rank( Q0 ) =2. So the given
0 1

system is completely observable.

Example 1.12: Consider the system with state equation, determine the observability
property

0 1 0  0   x1 
x(k  1)  0 0 1  x(k )  0  u (k ) Y  [3 4 1]  x2 
0 2 3 1   x3 

Solution:
The observability matrix is

 C 
Q0   CF 
CF 2 

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.20

 EI6801- Computer control of Process
0 1 0 
CF  [3 4 1] 0 0 1   [0 1 1]
0 2 3

0 1 0 
CF  CF  F  [0 1 1] 0 0 1   [0 2  2]
2

0 2 3

 C  3 4 1 
Q0   CF   0 1 1  & Q0  0
CF 2  0 2 2

The rank of the matrix is not equal to 3, so the system is unobservable.

Example 1.13: Consider the state variable model
 X 1 (k  1)   0 1   X 1 (k )  0   X (k ) 
 X (k  1)    0.125 0.75  X (k )   1  u (k ) & Y (k )   0.5 1  1 
 2    2     X 2 (k ) 
Y z
1)Find the transfer function G(z)= and determine the poles of the transfer
R z
function.
2) Comment on the controllability and observability property of the given system
without making any further calculations.

Solution:
Y z
= G(z)= C  ZI- F  g
-1
1)
R z
Z
0  0 1   Z 1 
 ZI-F    
Z   0.125 0.75 0.125 Z  0.75
0
 

1  Z  0.75 1 
 ZI-F  2
-1

(Z  0.75Z+0.125)  0.125 Z 
Y z 1   Z  0.75 1  0  (Z  0.5)
  0.5 1      2
R z (Z  0.75Z +0.125) 
2
 0.125 Z  1  Z  0.75Z +0.125
The poles of the transfer function are at Z=0.5 & Z=0.25.
2) Due to cancellation of pole zero term, system is uncontrollable and unobservable.
JURY’S STABILITY TEST:

The Jury’s stability test is used to determine whether the roots of the
characteristic polynomial lie within a unit circle or not. It consists of two parts,

 Necessary condition for stability.

 Sufficient condition for stability.
Consider the general form of the characteristic polynomial ∆z.

Let, ∆z be the nth order characteristic polynomial of a sampled data control system
Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.21
 EI6801- Computer control of Process
∆z = an zn + an-1zn-1 + an-2zn-2 +………+ a2z2 + az + a0 = 0

Where, an > 0 and a1, a2, a3……an are constant co-efficient.

Necessary Condition

∆z (1) > 0 and

(-1) n ∆z (-1) > 0 (or) ∆z (-1)> 0 for n even

∆z (-1) < 0 for n odd

Sufficient Condition

For sufficient condition, we construct a table whose elements are based on

the coefficients of ∆z.

∆z = an zn + an-1zn-1 + an-2zn-2 + …….+ a2z2 + a1z + a0 = 0

Row z0 z1 z2 ......zn-2 zn-1 zn

1 a0 a1 a2 ......an-2 an-1 an

2 an an-1 an-2 ...... a2 a1 a0

3 b0 b1 b2 ..... bn-1 bn-2

4 bn-2 bn-1 b2 ...... b1 b0

5 c0 c1 c2 .....Cn-1 Cn-2

6 Cn-2 Cn-1 c2 ......c1 c0

….

2n-5 p0 p1 p2 p3

2n-4 p3 p2 p1 p0

2n-3 q0 q1 q2

The table consists of (2n-3) rows, where n is the order of the characteristic equation.
The jury’s stability table is formed using the following rules,

1. The first two rows of the table consists of the co-efficient of ∆(z),arranged in
ascending order of powers of z in row 1,and a reverse order in row 2.
2. All even numbered rows are simply the reverse of the immediately preceding
odd-numbered rows.

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.22

 EI6801- Computer control of Process
3. The elements for rows 3 through (2n-3) are given by the following
determinants.
The kth elements of row-3 is given by,

a0 an k
bk =
an ak

a0 an a0 an 1 a0 an 2
b0  ; b1  ; b2  and so on
an a0 an a1 an a2

The row-4 is formed by arranging the element of row-3 is reverse order.

The kth elements of row-5 is given by,

b0 bn 1k
ck 
bn 1 bk

b0 bn 1 b0 bn 2 b0 bn 3
c0  ; c1  ; c2  and so on
bn 1 b0 bn 1 b1 bn 1 b2

The row-6 is formed by arranging the elements of row-5 in reverse order.

The procedure is continued until the (2n-3) rd row is reached which will contain
exactly three elements. The elements of the first column of the table are used to
check the following (n-1) conditions. These (n-1) conditions are the sufficient
conditions for the stability of the system.

a0  an

b0  bn1

c0  cn2

…… q0  q2

If the necessary and sufficient conditions are satisfied then all the poles of the system
lies inside the unit circle in Z-plane and so the system is stable. If even one of the
conditions is not satisfied the system is unstable.

Example:1.14: Consider the characteristics polynomial ∆(z) = 2 z 4  7 z 3  10 z 2  4 z  1

Check the stability of the system.

Applying stability constraints,

i) ∆z(1) = 2  7  10  4  1  24  0, satisfied

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.23

 EI6801- Computer control of Process
ii) ∆z(-1) = 2  7  10  4  1  2  0, satisfied
Row z0 z1 z2 z3 z4

1 1 4 10 7 2

2 2 7 10 4 1

3 -3 -10 -10 -1

4 -1 -10 -10 -3

5 8 20 20

Now, (n-1)=3 constraints are checked.

a0  an , (i.e) 1  2 satisfied

b0  bn1 , (i.e) 3  1 ; satisfied

c0  cn2 , (i.e) 8  20 ; not satisfied

The system is therefore unstable.

Example 1.15: Using Jury’s stability criterion check if all the roots of the following
characteristic equations lie within the unit circle.

i) z3 -1.3z 2 -0.08z+0.24=0

∆z(1)= 1-1.3-0.08+0.24=-0.14<0 not satisfied.

Thus the roots doesn’t lie within the unit circle

ii) z4 -1.368z3 +0.4z 2 +0.08z+0.002= 0

∆z(1)= 1-1.368-0.4+0.08+0.002=0.114 >0

∆z(-1)= 1+1.368+0.4-0.08+0.002=2.69 >0

The Jury table is

row Z0 Z1 Z2 Z3 Z4
1 0.002 0.08 0.4 -1.368 1
2 1 -1.368 0.4 0.08 0.002
3 -1 1.368 -0.4 -0.083
4 -0.083 -0.4 1.368 -1
5 1 -1.4 0.51

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.24

 EI6801- Computer control of Process
a0  an , (i.e) 0.002  1 satisfied

b0  bn1 , (i.e) 1  0.083 ; satisfied

c0  cn2 , (i.e) 1  0.51 ; satisfied

Hence all the roots of the characteristic equation lie within the unit circle.

Example 1.16: Using jury stability criterion, determine the values of ‘k’ for which the
system described by the characteristic equation z3 -0.1z 2 +0.2kz -0.1k = 0 is stable.

Solution: a0  an (i.e) |0.1k| < |1|

0.1k < 1
k<10
The Jury table is

row Z0 Z1 Z2 Z3
1 -0.1k 0.2k -0.1 1
2 1 -0.1 0.2k -0.1k
3 0.01 k2-1 -0.02 k2 +0.1 0.01k-0.2k
The constraint is, b0  bn1

|0.01 k2-1| > |-0.19k|

-(0.01 k2 – 1)>0.19k

k2+19k-100 < 0

19  (19)2  400 19  27.6

k1,2  = = 4.3, -23.3
2 2

(k-4.3)(k+23.3) < 0

k < 4.3

The system is stable for 0 < k <4.3

Example 1.17: Using jury stability criterion, examine the stability of the system
described by the characteristic equation z3 -1.1z 2 - 0.1z+0.2= 0

Solution:

Applying stability constraints,

i) ∆z(1) = 1  1.1  0.1  0.2  0, not satisfied

ii) ∆z(-1) = 1  1.1  0.1  0.2  1.8  0, satisfied

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.25

 EI6801- Computer control of Process
The Jury table is

row Z0 Z1 Z2 Z3

1 0.2 -0.1 -1.1 1

2 1 -1.1 -0.1 0.2

3 -0.96 1.08 -0.12

The constraint is, b0  bn1

b0  b2 i.e., 0.96>0.12 this condition is satisfied.

The first condition indicates that, at least one root is at z=1. Therefore the system is
critically stable.

Example 1.18: Check the stability of the system described by the characteristics
polynomial z  z 5  2.6 z 4  0.56z 3  2.05z 2  0.0775z  0.35

Solution:

Applying stability constraints,

i) ∆z(1) = 1  2.6  0.56  2.05  0.0775  0.35  1.4175  0, satisfied

 1 z (1)   1 1  2.6  0.56  2.05  0.0775  0.35  0.3825  0, satisfied
5
ii)
The Jury table contains (2n-3) rows.

row Z0 Z1 Z2 Z3 Z4 Z5
1 0.35 0.0775 -2.05 -0.56 2.6 1
2 1 2.6 -0.56 -2.05 0.0775 0.35
3 -0.877 0.832 -0.16 0.832
4 0.832 -0.16 0.832 -0.877
5 0.077 -0.6 -0.55
6 -0.55 -0.6 0.077
7 -0.29 -0.376
Now, (n-1)=4 constraints are checked.

a0  an , (i.e) 0.35  1 satisfied

b0  bn1 , (i.e) 0.877  0.832 ; satisfied

c0  cn2 , (i.e) 0.077  0.55 ; not satisfied

The constraint 3 is violated hence the system is therefore unstable.

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.26

 EI6801- Computer control of Process
Example 1.19: Consider the system shown in figure below using Jury stability
criterion. Find the range of K>0 for which the system is stable.

Solution: For a system with sample & hold, the forward path transfer function is
given by

 k   k  1 0.5 0.5  
Gh0 G( z )  (1  z 1 ) Z .T  2   (1  z 1 ) Z .T   2   
 s ( s  2)  2  s s s  2  

k  Tz 0.5z 0.5z  k  3( z  e6 )  0.5( z  1)( z  e6 )  0.5( z  1) 2 

 (1  z 1 )     
2T 
2  
( z 1)
2  ( z  1) z  1 z  e 
2
( z  1)2 ( z  e6 ) 

k  3( z  0.0025)  0.5( z  1)( z  0.0025)  0.5( z  1)2  k  3( z  0.0025)  0.5( z  1)(1  0.0025) 
  2
2  ( z  1)( z  0.0025)   ( z  1)( z  0.0025) 


The characteristics equation of the sampled data system is, 1+ Gh0 G( z ) =0

2( z 1)( z  0.0025)  k[2.5z  0.5]  0

2[ z 2  1.0025z  0.0025]  2.5kz  0.5k  0

 ( z)  z 2  z(1.0025  1.25k )  0.0025  0.25k

(1)  1  1.25k  1.0025  0.0025  0.25k  0

(i.e) 1.5k > 0, k>0

(1)  1  1.25k  1.0025  0.0025  0.25k  0

2.005-k > 0, hence k < 2.005

The system is found to be stable, for 0 < k < 2.005

Example 1.20: Investigate the stability of the system for sampling period T=0.5 and
T=1sec.

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.27

 EI6801- Computer control of Process

Solution:

The characteristic equation of the system is, 1+  G hoGH(s)  0

1- e-ST  k 
  G ho GH(s) =    
 S  S(S +1)  

Applying partial fraction expansion,

1 A B C
 2  
2
S (S +1) S S S 1

Solving for constants, A=1,B=-1,C=1

 1 1 1  -1  z 
  G ho GH(s) = k 1  z -1    2  
Tz z
 = k(1  z )   + T 
S S S 1   (z  1) z  1 z  e 
2

 T ( z  eT )  ( z  1)( z  eT )  ( z  1) 2 

  G ho GH(s)  k ( z  1)  
 ( z  1)2 ( z  eT ) 

Case i) T=0.5 sec

 0.5  z  e0.5    z  e0.5   z  1   z  12 

  G ho GH(s) = k  
  z  1  z  e0.5  
 

 0.5  z  0.61   z  0.61 z  1   z 2  2 z  1 

  G ho GH(s) = k  
  z  1 z  0.61 

 0.11z + 0.085 
  G ho GH(s) = k  
  z  1 z  0.61 

 0.11z + 0.085 
The characteristic equation of the system is, 1+ k  0
  z  1 z  0.61 

The poles of the closed loop system lie inside the unit circle, and the critical value of
K is found from magnitude criterion.

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.28

 EI6801- Computer control of Process
 0.11z+0.085 
k 1

  z 1 z  0.61  z 1

 0.11+ 0.085 
k  1  k  128.8
  1  1 1  0.61 

The system is found to be stable, for 0  k  128.8

Case ii) T=1 sec

  z  e1    z  e1   z  1   z  12 

  G ho GH(s) = k  
  z  1  z  e1  
 

  z  0.368   z  0.368 z  1   z 2  2 z  1   0.368z + 0.264 

  G ho GH(s) = k  =k 
  z  1 z  0.368    z  1 z  0.368  

 0.368z + 0.264 
The characteristic equation of the system is, 1+ k  0
  z  1 z  0.368  

The poles of the closed loop system lie inside the unit circle, and the critical value of
K is found from magnitude criterion.

 0.368z+0.264 
k 1

  z 1 z  0.368  z  1

 0.368 + 0.264 
k  1  k  26.33
  1  1 1  0.368  

The system is found to be stable, for 0  k  26.33

It is clear that, as the sampling rate increases the margin of stability decreases.

Schur-Cohn stability Test:

This method provides an analytical method for determining the absolute stability of
sampled data control systems. The characteristic equation for the sampled data
system is, 1  A( z )  F ( z)  0

Generally A(z) is the open loop transfer function and will be in the form of a ratio of
two polynomials.

F ( z)  a0  a1 z  a2 z 2  .......  an z n  0

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.29

 EI6801- Computer control of Process
The first step in determining the stability is to write the coefficients of ak in the
determinant form.  k is a determinant that has 2k rows and 2k columns. The Schur-
Cohn criteria states that all the roots of the characteristic equation lie inside the unit
circle if the following conditions are met.

k  0 for k odd
k  0 for k even

The determinant for few values of k is given below:

For k=1, 1  2  2 determinant.

a0 an
1 
an a0

For k=2, 2  4  4 determinant.

a0 0 an an 1
a1 a0 0 an
2 
an 0 a0 a1
an 1 an 0 a0

For k=3, 3  6  6 determinant.

a0 0 0 a3 a2 a1
a1 a0 0 0 a3 a2
a a1 a0 0 0 a3
3  2
a3 0 0 a0 a1 a2
a2 a3 0 0 a0 a1
a1 a2 a3 0 0 a0

Example 1.21: Determine the stability of the sampled data control system whose
 z 
open loop transfer function is given by A( z ) =  
  2.45z +1 2.45 z  1 

Solution:

The characteristic equation for the sampled data system is, 1  A( z )  F ( z)  0

z
1 0
 2.45z +1 2.45 z  1
F ( z )  (2.45z) 2  z  1  6 z 2  z  1  0
Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.30
 EI6801- Computer control of Process
Here the order of the characteristic equation is 2. Therefore, two determinants
1 &  2 are to be evaluated to determine stability.

a0 a2 1 6
1    35
a2 a0 6 1

a0 0 a2 a1 1 0 6 1
a1 a0 0 a2 1 1 0 6
2    1176
a2 0 a0 a1 6 0 1 1
a1 a2 0 a0 1 6 0 1

Since, 1  0 and 2  0 the system is stable.

If F ( z ) is a quadratic polynomial with real coefficients and the coefficient of z 2 is

unity, the Schur-Cohn criterion can be simplified. The necessary and sufficient
conditions that the roots of the characteristic equation lie inside the unit circle in the
Z plane are, F (0)  1; F (1)  0; F (1)  0

OPEN LOOP SAMPLE DATA SYSTEM:

y (t )  1  e ST 
  G p ( s)
r (t )  s 

Y ( z)  1  e ST   1  G p ( s) 
    G p ( s)   (1  Z )  
R( z )  s    S 

Example 1.22: Find the response of the system to a unit impulse input. [ Assume T=1
sec]

Solution:

Y ( z)  1  e ST   1   1  1 
       (1  Z )  
R( z )  s   S  1    S ( S  1) 

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.31

 EI6801- Computer control of Process
1
Applying Partial fraction equation to
S ( S  1)

1 A B
 
S ( S  1) S S  1

1  A(S  1)  BS

Solving for constants, A=1,B=-1

 1  t
L1  e
Y ( z) 1
 (1  z 1 )  
1   z
 (1  z 1 )  
z   S 1 

 S S  1  z  1 z  e 
1

  e T   
R( z ) n z
  z  eT
Y ( z)  z  e1  z  1  1  e1 0.632
 ( z  1)  1 
 1

R( z )  ( z  1)( z  e )  z  e z  0.368
r (n)   (n)  R( z)  1
0.632  1 
Y ( z)  z 1  n 1
  a u (n  1)
z  0.368  za

 y(n)  0.632(0.368)n1 u(n  1)

CLOSED LOOP SAMPLE DATA SYSTEM:

C ( z )   Gho (S )G(S ) E( z)  1

B( z )   Gho (s)G(s) H (s) E ( z )

E ( z )  R( z )  B( z )  R( z )   GhoGH (s) E ( z )

E ( z )  1   GhoGH (s)  R( z )

R( z )
E( z)   (2)
1   GhoGH ( S )

Substituting equation (2) in (1)

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.32

 EI6801- Computer control of Process
C(z)   G ho G(s)
=
R(z) 1+  G ho GH(s)

Example 1.23: Find the transfer function of the closed loop SDCS shown in fig.

Solution:

C(z)   G ho G(s)
=
R(z) 1+  G ho GH(s)

1- e-ST  1    1   z z 
  = 1- z   
0.632
  G ho G(s) =   
-1
 = (1- z -1 )  - -1 
=
 s  s +1    s(s +1)   z -1 z - e  z - 0.368

1- e-ST  1    1 
  G ho GH(s) =      1- z    2
-1
 
 s  s(s +1)    s (s +1) 

Applying partial fraction expansion,

1 A B C
  2
2
s (s +1) S S S 1

Solving for constants, A=1,B=-1,C=1

1  e-ST  1   1 1 1  -1  z 
   1  z    2  
z z
  G ho GH(s) =   
-1
 = (1  z )   + 
 s  s(s +1)   s s s 1  (z  1)
2
z  1 z  e-1 

 ( z  0.368)  ( z  1)( z  0.368)  ( z  1)2 

  G ho GH(s)  ( z  1)  
 ( z  1)2 ( z  0.368) 

( z  0.368)  ( z  z  0.368 z  0.368)  z 2  1  2 z 2  0.368  0.368 z  1

2

  G ho GH(s)  
( z  1)( z  0.368) ( z  1)( z  0.368)
0.368 z  0.264
  G ho GH(s) 
( z  1)( z  0.368)

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.33

 EI6801- Computer control of Process
0.632
C ( z) z  0.368 (0.632)( z  1)
 
R( z ) 1  (0.368 z  0.264) ( z  1)( z  0.368)  0.368 z  0.264
( z  1)( z  0.368)

C(z) 0.632z - 0.632

=
R(z) z 2 + z + 0.632

Example 1.24: Find the transfer function of the closed loop sample data system
shown in Fig below.

Solution:

C(z)   G ho G(s)
=
R(z) 1+   G ho G(s)

1- e-ST    
  = 1- z   
10 10
  G ho G(s) =   
-1
 
 s   s +1 s + 2     s  s +1 s + 2   

Applying partial fraction expansion,

 10  A B C
    
 s  s +1 s + 2   S S  1 S  2

Solving for constants, A=5, B=-10,C=5

 5 10 5  -1  5z 5z 
  G ho G(s) = 1- z -1    
10z
    (1- z )   + 
 s s  1 s  2    z -1 z - e z - e-2 
-1

 5( z  0.368)( z  0.135)  10( z  1)( z  0.135)  5( z  1)( z  0.368) 

  G ho G(s)  ( z  1)  
 ( z  1)( z  0.135)( z  0.368) 

4.3z  0.74
  G ho G(s) 
( z  0.135)( z  0.368)

4.3z  0.74
C ( z) ( z  0.135)( z  0.368) 4.3z  0.74
 
R( z ) 1  4.3z  0.74 ( z  0.135)( z  0.368)  4.3z  0.74
( z  0.135)( z  0.368)

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.34

 EI6801- Computer control of Process
C(z) 4.3z + 0.74
= 2
R(z) z  0.5z + 0.05

Example 1.25: Investigate the stability of the system for sampling period T=0.5 and
T=1sec.

Solution:

The characteristic equation of the system is, 1+  G hoGH(s)  0

1- e-ST  k  
  G ho GH(s) =    
 S  S(S +1)  

Applying partial fraction expansion,

1 A B C
 2  
2
S (S +1) S S S 1

Solving for constants, A=1,B=-1,C=1

 1 1 1  -1  z 
  G ho GH(s) = k 1  z -1    2  
Tz z
 = k(1  z )   + T 
S S S 1   (z  1) z  1 z  e 
2

 T ( z  eT )  ( z  1)( z  eT )  ( z  1) 2 

  G ho GH(s)  k ( z  1)  
 ( z  1)2 ( z  eT ) 

Case i) T=0.5 sec

 0.5  z  e0.5    z  e0.5   z  1   z  12 

  G ho GH(s) = k  
  z  1  z  e0.5  
 

 0.5  z  0.61   z  0.61 z  1   z 2  2 z  1 

  G ho GH(s) = k  
  z  1 z  0.61 

 0.11z + 0.085 
  G ho GH(s) = k  
  z  1 z  0.61 

 0.11z + 0.085 
The characteristic equation of the system is, 1+ k  0
  z  1 z  0.61 

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.35

 EI6801- Computer control of Process
The poles of the closed loop system lie inside the unit circle, and the critical value of
K is found from magnitude criterion.

 0.11z+0.085 
k 1

  z 1 z  0.61  z 1

 0.11+ 0.085 
k  1  k  128.8
  1  1 1  0.61 

The system is found to be stable, for 0  k  128.8

Case ii) T=1 sec

  z  e1    z  e1   z  1   z  12 

  G ho GH(s) = k  
  z  1  z  e1  
 
  z  0.368   z  0.368 z  1   z 2  2 z  1   0.368z + 0.264 
  G ho GH(s) = k  =k 
  z  1 z  0.368    z  1 z  0.368  
 0.368z + 0.264 
The characteristic equation of the system is, 1+ k  0
  z  1 z  0.368  
The poles of the closed loop system lie inside the unit circle, and the critical value of
K is found from magnitude criterion.

 0.368z+0.264 
k 1

  z 1 z  0.368  z  1

 0.368 + 0.264 
k  1  k  26.33
  1  1 1  0.368  
The system is found to be stable, for 0  k  26.33
It is clear that, as the sampling rate increases the margin of stability decreases.

Example 1.26: Determine the response of the closed loop sample data system shown
in Fig to unit step input.

Solution:

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.36

 EI6801- Computer control of Process
  RG(s)
C(z) =
1 +   GG ho (s)

 1  1 1   z z  0.632
  RG(s) =    =    =  -1 
=
 S  S +1   S (s +1)   z  1 z  e  z  0.368

 1  e ST   1    1 
  GG ho (s) =   
1
  = (1  Z )  
 s   S  1    S (S  1) 

 1  t
L1  e
1
 GG ho (s)  (1  z 1 )  
1   z
 (1  z 1 )  
z   S 1 

 S S  1  z  1 z  e 
1

  e T   
n z
  z  e T
 z  e1  z  1  1  e1 0.632
 GG ho (s)  ( z  1)  1 
 1

 ( z  1)( z  e )  z  e z  0.368

0.632
  RG(s)
 z  0.368 
0.632 C(z) 0.632
C(z) =  
1+   GG ho (s) 1+ 0.632 z  0.264 z z  z  0.264 
z  0.368

0.632
Applying partial fraction expansion to
z  z  0.264 
0.632 A B
 
z  z  0.264  z  z  0.264 
Solving for constants A and B we get, A=-2.4, B=2.4
z 2.4 z
C ( Z )  2.4 
z  z  0.264 

c(n)  2.4 (n)  2.4  0.264  u(n) 

n

 
Pole placement or State feedback controller Design:
In pole placement design method all the closed loop poles may be placed in
desired location. This design technique starts with the determination of desired
closed-loop poles to satisfy transient response requirements. By choosing an
appropriate gain matrix K for state feedback, it is possible to force the system to have
closed loop poles at the desired locations, provided that the original system is
completely state controllable. In this design technique, it is assumed that all state
variables are measurable and are available for feedback.

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.37

 EI6801- Computer control of Process

Consider a nth order SISO system with state feedback control law,
u( K )   K X ( K )

The state equation of the system with state variable feedback is given by
 X ( K  1)  FX ( K )  gu( K )  FX ( K )  g K X ( K )  ( F  g K ) X ( K )
The Eigen values of ( F  gK ) can be arbitrarily placed in the complex plane by

choosing K suitably.
State feedback gain matrix determination:
i) Check for state controllability
ii) Determine the desired characteristic polynomial from the given desired pole
locations
iii) Determine the matrix  ( F ) by comparing the coefficients of the desired
characteristic polynomial, Where  ( F )  F n  a1F n1  a2 F n2  ..........  an1F  an I

iii) Determine state feedback gain matrix k, using Ackermann’s formula as given
by k  [ 0 0 ………0 1] U 1 ( F )
Where U= [g Fg………Fn-1g]

Example 1.27: A discrete time regulator system has the plant

0 1 0 0 
x(k  1)   0 0 1  x(k )  0  u (k )
 
 4 2 1 1 
Design a state feedback controller which will place the closed loop poles at
0.5  j 0.5,0 .
Solution:
The desired characteristics polynomial is
z( z  (0.5  0.5 j ))( z  (0.5  0.5 j))  z 3  z 2  0.5z  0

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.38

 EI6801- Computer control of Process
z n  a1 z n 1  a2 z n 2  ...........an 1 z  an  0
 a1  1; a2  0.5; a3  0
Using Ackermann’s formula k  [ 0 0 ………0 1] U 1 ( F )
 ( F )  F n  a1F n1  a2 F n2  ..........  an1F  an I
U   g Fg F 2 g 
K   0 0 1U 1 ( F )
0   0 1 0  0  0 
g  0  ;  Fg    0 0 1  0   1 
1   4 2 1 1   1
 0 1 0  0   1 
 F g   F ( Fg )   0 0 1   1    1
2
    
 4 2 1  1  1
1
 0 1 1  2 1 1
T T T
0 0 1  2 1 1  2 1 1 
U  0 1 1   1 1 0   (1)  1 1 0   1 1 0   1 1 0 
 
1 1      
1
1 1 1  1 0 0   1 0 0  1 0 0  1 0 0 
 ( F )  F 3  a1 F 2  a2 F  a3 I
 4 2 1
F  F [ F ]   4 2 1
3 2

 4 6 1
0 0 1
a1 F   4
2
2 1
 4 2 1
0 1 0
1
a2 F   0 0 1 
2
 4 2 1
 4 2 1  0 0 1   0 0.5 0   4 1.5 0 
 ( F )   4 2 1   4 2 1   0 0 0.5    0 4 1.5
      
 4 6 1  4 2 1  2 1 0.5  6 3 2.5
K  0 0 1U 1 ( F )
 2 1 1   4 1.5 0 
  
K   0 0 1 1 1 0   0 4 1.5
1 0 0  6 3 2.5
K   4 1.5 0

Aliter:
The desired characteristics polynomial is
z( z  (0.5  0.5 j))( z  (0.5  0.5 j))  z 3  z 2  0.5 z  0  (1)

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.39

 EI6801- Computer control of Process
The characteristics equation is ZI  F  gK )  0

 z 0 0  0 1 0  0
0 z 0    0 0 1   0  K
      1 K2 K3   0
0 0 z   4 2 1 1 
 z 1 0 
 0 z 1   0

 4  K1 2  K 2 z  1  K 3 
Z 3  z 2 (1  K3 )  (2  K2 ) z  (4  K1 )  0  (2)
Comparing equation (1) and (2)
(4  K1 )  0  K1  4
(2  K 2 )  0.5  K 2  1.5
(1  K3 )  1  K3  0
Example 1.28: Consider the system defined by
 0 1 0
 X (k  1)     X (k )     u (k )
 0.16 1 1 
Y (k )  1 0 X (k )
Design a state feedback controller which will place the closed loop poles at
1  j 2 .
Solution:
The desired characteristics polynomial is
( z  (1  2 j ))( z  (1  2 j))  z 2  2 z  5  0  (1)
The characteristics equation is ZI  F  gK )  0
 z 0  0 1  0 
0 z    0.16 1  1   k1 k2  0
     
 z 1   0 0
0.16 z  1   k k2 
0
   1
 z 1 
 k  0.16 z  1  k   0
 1 2

z 2  (1  k2 ) z  0.16  k1  0  (2)
Comparing equation (1) and (2)

(0.16  k1 )  5  k1  4.84
(1  k2 )  2  k2  1

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.40

 EI6801- Computer control of Process

State observer/State Estimator design:

State observer is a system in the control system that performs an estimation of
the state variables based on the measurements of the output and control variables.
 Full order state observer  observes all n state variables.
 Minimum order state observer  observes only un measurable state variables.
 Reduced state observer  observes un measurable and measurable state
variables.

The observer should have the same state equations as the original system &
design criterion should be to minimize the difference between the system output
y=[c][x] & the output ŷ =[c][ x̂ ] as constructed by the observed vector x̂ . This is
equivalent to minimization of x  xˆ . Since X is inaccessible an attempt is made to
minimize ( y  yˆ ) The difference ( y  yˆ ) is multiplied by n x 1 vector m & fed into the
input of the integrator of the observer. The problem is to design’ m’ so that ( y  yˆ ) is
minimized.

We know that, X  k  1  F X  k   g u  k  
1
Let Xˆ (k ) is theobsrved state

Refering to Fig.,
Xˆ (k  1)  FXˆ (k )  gU (k )  m  y(k )  yˆ (k )

Xˆ (k  1)  FXˆ (k )  gU (k )  m c X (k )  c Xˆ (k ) 

Xˆ (k  1)   F  mc  Xˆ (k )  gU (k )   mc X (k )  
2

Equation (1)-(2) results in,

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.41

 EI6801- Computer control of Process
X  k  1  Xˆ (k  1)   F  mc   X (k )  Xˆ (k ) 

let X (k )  Xˆ (k )  X (k )
Then X  k  1   F  mc  X (k )
The vector X  k  1 can be made to decay fast if the Eigen value of the matrix
 F  mc  are properly selected.
Procedure:
i) Check for state observability
ii) Determine the characteristic polynomial of the system with state observer,
given by ZI  ( F  mC )  0
Wher m   m1 m2 .... mn 
T

Let ZI  ( F  mC )  Z n  b1 Z n1  .............  bn1Z  bn

The coefficients of this polynomial (ie) b1, b2…………bn will be the functions of m1,
m2……mn..
iii) Determine the desired characteristic polynomial from the specified Eigen
values.
Let the specified Eigen values be 1,2 ..........n Now, the desired characteristic
polynomial is, (  1 )(  2 )........(  n )   n  a1 n1  .........  an1  an

iv) The coefficients of polynomials are equated.

b1  a1 , b2  a2 , ........, bn  an , On solving, the elements m1, m2 ,....., mn of state
observer gain matrix ‘m’ can be obtained.
(Or) Using the Ackermann’s formula, the state observer gain matrix G can be
calculated.
1  0 
 C   
 CF  0 
  .
m   ( A)  .   
  .
 .  0 
CF n 1   
1 
Where  ( F )  F n  a1 F n1  ........  an1 F  an I .
Example 1.29: Consider the system
X (k  1)  FX (k )  g U (k )
y (k )  CX (k )
0 0.16  0 
where F  
1 
g  1  C   0 1
1  
Design a full order observer which places the closed loop characteristic roots at
0.5  0.5 j

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.42

 EI6801- Computer control of Process
Desired observer feedback gain matrix
characteristic equation ZI  F  mC  0
m 
let m   1 
 m2 
1 0  0 0.16   m1 
  0 1  0
1   m2 
Z
 0 1  1
z m1  0.16
0
1 z  1  m2
z ( z  1  m2 )  m1  0.16  0
z 2  z (1  m2 )  m1  0.16  0 1
The desired characteristic equation is z 2  z  0.5  0 2
Comparing 1 & 2,
1  m2  1 m1  0.16  0.5
m2  2 m1  0.34

0.34
The observer gain matrix is, m   
 2 
Dead beat observer design:
Consider a completely controllable and observable SISO system of order n, described
by the equations,
X  k  1  F X  k   g u  k  
1
Y k   C X k   D u k  
2

With state feedback control law,

U ( k )  k X ( k )
Now equation 1 becomes,
X  k  1   F  gk  X  k 
The system has the characteristic equation,
ZI   F  gk   0 
3
The controller design consists of picking the gains k, so that equation (3) matches the
desired characteristic equation Z n  0
Example 1.30: A discrete time regulator system has the plant
 0.16 2.16   1
x(k  1)    x(k )    u (k )
 0.16 1.16 1

y(k )  1 1 X (k )

a)Design a state feedback controller which will place the closed loop poles at
0.6  j 0.4 .
b) Design a prediction observer for dead beat response.
Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.43
 EI6801- Computer control of Process
Solution:
state feedback controller design:
The desired characteristics polynomial is
( z  0.6  0.4 j ))( z  0.6  0.4 j ))  0
z 2  1.2 z  0.52  0
The given plant model is converted into controllable canonical form.
 1 2 
The controllability matrix is, Qc   g Fg    
 1 1
The determinant of matrix Qc is not equal to zero, so the system can be transformed
into CCF.

 Z  0.16 2.16 
The characteristic equation of A is, ZI  F     Z 2  Z  0.16  0
 0.16 Z  1.16

The coefficients of the characteristic equation are identified as a0  0.16, a1  1

Now,

 a 1  1 1 
M  1  
 1 0 1 0
1 1
P  Qc M   
0 1 
The CCF model is given by,

 1 0  2 1 1 1  0 1
F  P 1FP         
 1 1   1 1  0 1   0.16 1

0 
g  P 1 g   
1 

The characteristics equation is ZI  F  gK )  0

 z 0  0 1  0 
0 z    0.16 1  1   k1 k2   0
     
 z 1   0 0
0.16 z  1   k k2 
0
   1
 z 1 
 k  0.16 z  1  k   0
 1 2

z 2  (1  k2 ) z  0.16  k1  0  (2)
Comparing equation (1) and (2)

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.44

 EI6801- Computer control of Process
(0.16  k1 )  0.52  k1  0.36
(1  k2 )  1.2  k2  2.2

K  0.36 2.2
K  KP1   2.56 2.2

Design of prediction observer for dead beat response:

The desired characteristics polynomial is z 2  0
The characteristics equation is ZI  F  mC )  0

 z 0  0 1   m1 
0 z    0.16 1   m  1 0  0
     2
 z 1   m1 0
0.16 z  1   m 0
0
   2
 z  m1 1 
 m  0.16 z  1  0
 2 
z 2  (1  m1 ) z  m1  0.16  m2  0  (2)
Comparing equation (1) and (2)

(1  m1 )  0  m1  1
m1  0.16  m2  0  m2  0.84

Current Observer design:

The prediction observer given arrives at the state estimate Xˆ (k ) after receiving
measurements up through y(k  1) . Hence the control does not utilize the
information on the current output y(k). For higher order systems controlled with a
slow computer, or any time the sample rates are fast compared to the computation
time, this delay between making a measurement and using it in control law may be a
blessing. In many systems, however, the computation time is quite short compared to
the sample period and the control based on prediction observer may not be as
accurate as it could be.
An alternative formulation of the state observer is to use y(k) to obtain the state
estimate. This can be done by separating the estimation process into two steps. In the
first step X (k  1) is determined. In the second step, y(k  1) is used to
improve X (k  1) . The state observer based on this formulation is called the current
observer. The current observer equations are,
X  k  1  F Xˆ  k   g u  k  1
Xˆ (k  1)  FXˆ (k )  gU (k )  m  y(k  1)  yˆ (k  1)

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.45

 EI6801- Computer control of Process
Xˆ (k  1)  FXˆ (k )  gU (k )  m C X (k  1)  C X  k  1  (2)

X  k  1  Xˆ (k  1)  F X  k   g u  k    FXˆ (k )  gu (k )  m C X (k  1)  C X  k  1  

X  k  1   F  mCF   X (k )  Xˆ (k )    F  mCF  X (k )

The vector X  k  1 can be made to decay fast if the Eigen value of the matrix
 F  mcF  are properly selected.
Example 1.31: Consider the system defined by
 0 1 0
 X (k  1)     X (k )     u (k )
 0.16 1 1 
Y (k )  1 1 X (k )
Design a current observer for the system; the response to the initial observer error
is required to be dead beat.
Solution:
The characteristics equation is ZI  ( F  mCF )  0

 z 0  0 1   m1   0 1
0 z    0.16 1   m  1 1  0.16 1  0
     2  
 z 1   m1 0  1 1   0 1
0.16 z  1   m 
0  0 0  0.16 1
   0
   2
 z 1   m1 m1   0 1
0.16 z  1   m 
m2   0.16 1
 0
   2
 z  0.16m1 1 
0.16  0.16m 0
 2 z  1
( z  0.16m1 )( z  1)  (0.16  0.16m2 )  0
z 2  0.16m1 z  z  0.16m1  0.16  0.16m2  0
z 2  (0.16m1  1) z  0.16m1  0.16  0.16m2  0
The desired characteristics polynomial is z 2  0
(0.16m1  1)  0
1
m1   6.25
0.16
0.16m1  0.16  0.16m2  0
0.16(6.25)  0.16  0.16m2  0
0.16  1  0.16m2
m2  5.25

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.46

 EI6801- Computer control of Process

Design of reduced order state observer:

This observer estimates only those states that are not measured. Let us
consider the case of only one measurement (i.e.,) state variable x1 is measured. The
output equation is then given by, y(t)= x(k)=x1(k)
To derive the reduced order observer, the state vector x(t) is partitioned into
two parts namely x1 which is directly measured and the other part is xe representing
the state variables that need to be estimated.

 x1 (t ) 
i.e., x(t )   
 xe (t ) 
The complete description of system is given by,

 x1 (k  1)   f11 f1e   x1   g1 
 x (k  1)    f 
f ee   xe   ge 
u
 e   e1
 x1 
y  1 0  
 xe 
The dynamics of the unmeasured state variables are given by,
xe (k  1)  fe1 x1  fee xe  geu  f ee xe  f e1x1  geu  (1)

Since x1=y, the measured dynamics are given by,

x1 (k  1)  y(k  1)  f11 x1  f1e xe  g1u

y(k  1)  f11 x1  f1e xe  g1u

y(k  1)  f11 y  g1u  f1e xe

x(k  1)  Fx(k )  gu(k )  (2)

The following substitution is made in the original observer design equation to obtain
reduced order observer.
From (1) & (2)
x  xe

F  fee

gu  fe1 y  ge u

y(k )  y(k  1)  f11 y(k )  g1u(k )

C  f1e

Thus the reduced order observer equations are,

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.47

 EI6801- Computer control of Process
xˆ(k  1)  Fxˆ(k )  gu(k )  m[ y(k )  y(k  1)]

xˆe (k  1)  f ee xˆe (k )  f e1 y(k )  geu (k )  m[ y(k  1)  f11 y(k )  g1u (k )  f1e xˆe (k )]
    
input measurement

Let xe  xe  xˆe  (3)

By differentiating, xe (k  1)  xe (k  1)  xˆe (k  1)  (4)

Where, xe (k  1)  fee xe  f e1 x1  geu  (5)

Substituting (3) & (4) in (5), & simplifying, the error equation is,
xe (k  1)  ( f ee  mf1e ) xe (k )

The characteristics equation of the reduced order observer is ZI  ( fee  mf1e )  0

Example 1.32: Consider the system defined by

 0 1 0
 X (k  1)     X (k )     u (k )
 0.16 1 1 

Y (k )  1 1 X (k )

Design a state feedback controller so that the response system is dead beat.

Solution:
The desired characteristics polynomial is z 2  0  (1)
The characteristics equation is ZI  F  gK )  0
 z 0  0 1  0 
0 z    0.16 1  1   k1 k2  0
     
 z 1   0 0
0.16 z  1   k k2 
0
   1
 z 1 
 k  0.16 z  1  k   0
 1 2

z  (1  k2 ) z  0.16  k1  0
2
 (2)
Comparing equation (1) and (2)

(0.16  k1 )  0  k1  0.16
(1  k2 )  0  k2  1
Example 1.33: A discrete time regulator system has the plant
 1 1 0
x(k  1)    x( k )    u (k )
 0 2 1 

Design a state feedback controller which will place the closed loop poles at 0.5,0.6.

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.48

 EI6801- Computer control of Process
Solution:
state feedback controller design:
The desired characteristics polynomial is
( z  0.6)( z  0.5)  0
z 2  1.1z  0.3  0  (1)
The given plant model is converted into controllable canonical form.
0 1
The controllability matrix is, Qc   g Fg    
1 2
The determinant of matrix Qc is not equal to zero, so the system can be transformed
into CCF.

Z  1 1 
The characteristic equation of A is, ZI  F     Z 2  3Z  2  0
 0 Z  2

The coefficients of the characteristic equation are identified as a0  2, a1  3

Now,

 a 1  3 1 
M  1  
 1 0 1 0
 1 0
P  Qc M   
 1 1
The CCF model is given by,

 1 0  1 1  1 0  0 1 
F  P 1FP      
 1 1   0 2  1 1   2 3

0 
g  P 1 g   
1 

The characteristics equation is ZI  F  gK )  0

 z 0  0 1  0
0 z    2 3  1  k1 k2   0
     
 z 1   0 0 
 2 z  3   k k   0
   1 2
z 2  (3  k2 ) z  2  k1  0  (2)
Comparing equation (1) and (2)

(2  k1 )  0.3  k1  1.7
(3  k2 )  1.1  k2  4.1

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.49

 EI6801- Computer control of Process

K   1.7 4.1
K  KP1   2.4 4.1
Example 1.34: Consider the system defined by

1 1 0.5
 X (k  1)     X (k )     u (k )
0 1 1

Y (k )  1 0 X (k )

Find the estimator equations and the value of the gain so that the desired poles are
at 0.6  j 0.3 for

i) A prediction estimator
ii) A current estimator

Design of prediction estimator:

The desired characteristics polynomial is
 z  0.6  0.32  z 2  1.2 z  0.45  0  (1)
2

The characteristics equation is ZI  F  mC )  0

 z 0 1 1  m1 
0 z   0 1   m  1 0  0
     2
 z  1 1   m1 0
 0
 0
 z  1  m2 0

 z  1  m1 1 
 m 0
 2 z  1
z 2  (m1  2) z  m1  1  m2  0  (2)
Comparing equation (1) and (2)
(2  m1 )  1.2  m1  3.2
m1  1  m2  0.45  m2  2.65
Design of current estimator:
The desired characteristics polynomial is
 z  0.6  0.32  z 2  1.2 z  0.45  0  (1)
2

The characteristics equation of the closed loop system is ZI  ( F  mCF )  0

 z 0 1 1  m1  1 1
0 z   0 1   m  1 0 0 1  0
     2  
 z  1 1   m1 0 1 1
 0
 0
 z  1  m2 0 0 1

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.50

 EI6801- Computer control of Process
 z  1 1   m1 m1 
 0
 0
 z  1  m2 m2 

 z  1  m1 1  m1 
 m 0
 2 z  1  m2 
z 2  (m1  m2  2) z  1  m1  0
1  m1  0.45  m1  0.55
m1  m2  2  1.2  m2  2.65

SOLUTIONS TO ASSIGNMENT PROBLEMS

1. The state diagram of a linear system is shown in Fig. Assign the state variables and
write the dynamic equations of the system in matrix form.

Solution:

Let X 1 be state variable assigned to the output of first delay element, X 2 be state

variable assigned to the output of second delay element and X 3 be state variable

assigned to the output of third delay element. Referring to Fig…,

X1 (k  1)  0.5 X1 (k )  0.25 X 2 (k )  2 X 3 (k )  U (k )

X 2 (k  1)  0.5 X 2 (k )  X 3 (k )  U (k )

X 3 (k  1)  3 X 2 (k )  0.33 X 3 (k )  2U (k )

Y (k )  5 X1 (k )  6 X 2 (k )  7 X 3 (k )  8U (k )

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.51

 EI6801- Computer control of Process
Putting the above equations in matrix form,

0.5 0.25 2 1 
X (k  1)  0  0.5  1  X (k )   1  U (k )
 
0 3 0.33   20

 X1 ( K ) 
 
Y ( K )  5 6  7   X 2  K    8U ( K )
X K 
 3  

2.Obtain the companion form realizations for the following transfer

Y  z  2z3 +2z 2 +17z+2
function = 3
u z z  z2  z  2

Solution:

Y  z  2z3 +2z 2 +17z+2

= 3
u z z  z2  z  2

Controllable phase variable form:

 X 1 (k  1)  0 1 0   X 1 ( K )  0 
 X (k  1)   0 0 1   X K   0  U ( K )
 2     2    
 X 3 (k  1)   2 1 1   X  K   1 
 3 

Elements of C matrix:

b0  2; b1  2; b2  17; b3  2; a1  1; a2  1; a3  2

b3  a3b0  2  (2)*(2)  2
b2  a2b0  15
b1  a1b0  4

 X1 ( K ) 
 
Y ( K )   2 15 4   X 2  K    2 U ( K )
 
 X3  K 

X ( k  1)
1
 X 2 (k )
X ( k  1)
2
 X 3 (k )
X ( k  1)
3
 2 X1 (k )  X 2 (k )  X 3 (k )  U (k )
Y (k )  2 X1 (k )  15 X 2 (k )  4 X 3 (k )  2U (k )

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.52

 EI6801- Computer control of Process

Observable phase variable form:

 X 1 (k  1)  0 0 2   X 1 ( K )   2 
 X (k  1)   1     
 2   0 1   X 2  K    15  U ( K )
 X 3 (k  1)  0    4 
 X3  K   
1 1

 X1 ( K ) 
 
Y ( K )   0 0 1   X 2  K    2U ( K )
X K 
 3  

X ( k  1)  2 X 3 ( k )  2 U ( k )
1

X ( k  1)  X1 (k )  X 3 (k )  15U (k )
2
X ( k  1)  X 2 ( k )  X ( k )  4 U ( k )
3 3
Y (k )  X 3 (k )  2U (k )

3.A discrete time system has the state equation given by,

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.53

 EI6801- Computer control of Process
0 1 
X (k  1)    X (k )
 10  7 

Find state transition matrix (k ) using Cayley Hamilton theorem.

Solution:

 0  0 1   1 
The Eigen values are,  I  F          2  7  10  0
 0    10  7  10   7 

  2, 5

Since ‘F’ is second order, g     0 I  1

The coefficients 0 and 1 are evaluated using the following equations

f  1   1k  g  1 

f  2   2k  g  2 

f  2    2   0  21
k

f  5   5  0  51

k

Solving above two equations we get,

1  0.33  2   0.33  5

k k

0  1.66  2  0.66  5

k k

The state transition matrix   k   0 I  1F

1 0  0 1 
  k   1.66  2   0.66  5    0.33  2   0.33  5   10  7 
k k k k

0 1   

 1.66  2 k  0.66  5k 0.33  2   0.33  5  

k k

 k    
 3.33  2   3.33  5  0.65  2   1.65  5  
k k k k

4. Consider a DTS with the characteristic polynomial,

z  z 3  0.2 z 2  0.2kz  0.2k .Determine the values of ‘k’ for which the system is stable.

Solution: | a0| < | an| (i.e) |- 0.2k| < |1|

0.2k < 1
k<5

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.54

 EI6801- Computer control of Process
The Jury table is

row Z0 Z1 Z2 Z3

1 -0.2k 0.2k -0.2 1

2 1 -0.2 0.2k -0.2k

3 0.04 k2-1 -0.04 k2 +0.2 0.04k-0.2k

The constraint is, | bn-1 | < | b0 |

|0.04 k2-1| > |-0.16k|

- (0.04 k2 – 1)>0.16k

k2+4k-25 < 0

4  16  100 4  10.8
k1,2    7.38,3.4
2 2

k < 3.4

The system is stable for 0 < k <3.4

5. Using Jury’s stability test, check if all the roots of the following characteristic
equation lie within the unit circle. z3 + 2.1z2 +1.44z + 0.32 = 0

Solution:

∆z(1)= 1+2.1+1.44+0.32 = 4.86 >0 satisfied.

(-1)∆z(-1)= 1+2.1-1.44+0.32 = 1.98 >0 satisfied.

The Jury table is

row Z0 Z1 Z2 Z3
1 0.32 1.44 2.1 1
2 1 2.1 1.44 0.32
3 -0.898 -1.64 -0.78
(n-1) constraints are checked.

a0  an , (i.e) 0.32  1 satisfied

b0  bn1 , (i.e) 0.898  0.78 ; satisfied

Thus the roots lie within the unit circle.

6. Using jury stability criterion, determine the values of ‘k’ for which the system
described by the characteristic equation z 3 -0.25z 2 -0.05z+k =0 is stable.
Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.55
 EI6801- Computer control of Process
Solution:

The Jury table is ,

row Z0 Z1 Z2 Z3
1 k - 0.05 -0.25 1
2 1 -0.25 -0.05 k
3 k2-1 -0.05 k + 0.25 -0.25k+0.05
(n-1) constraints are checked.

i) | an|<| a0| (i.e) | k| < |1|

ii) | bn-1 | < | b0 |
|k2-1| > |-0.25k+0.05|

k2 +0.25k – 1.05 > 0

0.25  (0.25)2  4.2

k1,2   1.16, 0.91
2

(k+1.16)(k-0.91) > 0

The system is stable for 0.91< k <1

7. Using Schur-Cohn’s stability test, check if all the roots of the following characteristic

equation lie within the unit circle. 2.661z 2 +2.508 z -0.8187 = 0

Solution:

F ( z )  2.661z 2 +2.508 z - 0.8187 = 0

Here the order of the characteristic equation is 2. Therefore, two determinants

1 &  2 are to be evaluated to determine stability.

a0 a2 0.8187 2.661
1    6.41
a2 a0 2.661 0.8187

a0 0 a2 a1 0.8187 0 2.661 2.508

a1 a0 0 a2 2.508 0.8187 0 2.661
2    35.1
a2 0 a0 a1 2.661 0 0.8187 2.508
a1 a2 0 a0 2.508 2.661 0 0.8187

Since, 1  0 and 2  0 the system is unstable.

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.56

 EI6801- Computer control of Process
8.Consider a plant defined by the following state model
0.5 1 0  1 4
x(k  1)   1 0 1  x(k )   0 0  u (k )
 
 0 0 0   3 2 

C  1 0 0; d  0 4

Design a prediction observer such that the observer poles are at 0.5  j 0.25

Solution:

The desired characteristics polynomial is

( z  0.5  0.25 j ))( z  0.5  0.25 j ))  0

z 3  z 2  0.3125z  0

The characteristics equation is ZI  F  mc)  0

 z 0 0  0.5 1 0   m1 
0 z 0   1 0 1    m  1 0 0  0
     2  
0 0 z   0 0 0  m3 

 z  m1  0.5 1 0 
 1 m z 1  0
 2

 m3 0 z 

Z 3  z 2 (m1  0.5)  (1  m2 ) z  m3  0  (2)

Comparing equation (1) and (2)

(m1  0.5)  1  m1  1.5

(1  m2 )  0.3125  m2  0.6875
m3  0  m3  0

9.Consider a plant defined by the following state model

 0 1 0 0 

x(k  1)   0 0 1  x(k )  0  u (k )

 0.5 0.2 1.1 1 

Determine the state feedback gain matrix ‘k’ such that the closed loop system will
exhibit the dead beat response.

Solution:

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.57

 EI6801- Computer control of Process
The desired characteristics polynomial is z 3  0  (1)

The characteristics equation is ZI  F  gK )  0

 z 0 0  0 1 0  0 
0 z 0    0 0 1   0   K1 K2 K3   0
  
0 0 z   0.5 0.2 1.1 1 
 z 1 0 
 0 z 1  0
 
0.5  K1 0.2  K 2 z  1.1  K 3 

Z 3  z 2 ( K3  1.1)  (0.2  K2 ) z  (0.5  K1 )  0  (2)

Comparing equation (1) and (2)

( K3  1.1)  0  K3  1.1
(0.2  K 2 )  0  K 2  0.2
(0.5  K1 )  0  K1  0.5

10. Consider a plant defined by the following state model

 0 1 0 0 

x(k  1)   0 0 1  x(k )  0  u (k )

 0.5 0.2 1.1 1 

Assuming that only y(k)=x2(k) is measurable, design a reduced order observer such
that the response to the observer error is dead beat. Give all relevant observer
equations.

Solution:

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.58

 EI6801- Computer control of Process

11. A regulator system has the plant with the transfer function

Y ( z) z 2

U ( z ) 1  0.8 z 1 1  0.2 z 1 

i) Obtain the plant model in controllable canonical form.

ii) Design a state feedback controller such that the closed loop poles are at
0.6  j 0.4
iii)Design a first order observer that estimates X2, given measurements of X1,
the response to the initial observer required is dead beat.
Solution:

Y ( z) z 2 z 2 1
  
U ( z ) 1  0.8 z 1  0.2 z   0.16 z  z  1  0.16  z  z 2 
1 1 2 1

i) State model in controllable form is,

 X 1 (k  1)   0 1   X 1 (k )   0 
 X (k  1)    0.16 1  X (k )   1  u (k )
 2    2   
y (k )  1 0 X (k )

ii) The desired characteristics polynomial is

( z  0.6  0.4 j ))( z  0.6  0.4 j ))  0  z 3 1.2 z  0.52  0

The characteristics equation is ZI  F  gK )  0

 z 0  0 1  0 
0 z    0.16 1  1   k1 k2   0
     

 z 1   0 0
0.16 z  1   k k2 
0
   1

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.59

 EI6801- Computer control of Process
 z 1 
 k  0.16 z  1  k   0
 1 2

z 2  (1  k2 ) z  0.16  k1  0  (2)

Comparing equation (1) and (2)

(0.16  k1 )  0.52  k1  0.36

(1  k2 )  1.2  k2  2.2

The characteristics equation is ZI  Fee  mf1e )  0

 0 1
F  
 0.16 1
Fee  1 f1e  1 fe1  0.16
Z  1  m)  0
The desired characteristics polynomial is z=0

1 m  0  m  1

xˆ2 (k  1)  Fee xˆ2 (k )  f e1 y(k )  g eu (k )  m  y (k  1)  f11 y  k   g1u (k )  f1e xˆ2 (k ) 

  1  m  xˆ2 (k )  0.16 y(k )  u (k )  my(k  1)

xˆ2 (k  1)  0.16 y(k )  u(k )  y(k  1)

UNIT-1 Two mark questions and Answers

1. Define state transition matrix.

 (k )  1  ZI  F  Z 
1
 

2. Define controllability of discrete time system.

For the linear system given by equation,

X (k  1)  F X (k )  gU (k )
y ( k )  c X (k )  d U (k )
If there exists an input u(k), which transfers the initial state X(0) = X 0 to the state X1 at
k=N, the state X0 is said to be controllable. If all initial states are controllable, the
1.
system is said to be completed controllable. Otherwise, the system is said to be
uncontrollable. The necessary and sufficient condition for the system to be
completely controllable is that the nxn controllability matrix,Qc = [g, Fg, F2g ….. Fn-
1g] has rank equal to n. i.e., rank(Qc) = n.

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.60

 EI6801- Computer control of Process

3.Define observability of discrete time system.

For the linear system given by equation,

X (k  1)  F X (k )  gU (k )
y ( k )  c X (k )  d U (k )
If it is possible to determine the state X(0) = X0 by the measurement of output Y(k)
over a finite time interval then the state X0 is said to be observable. If all initial
2.
conditions are observable, the system is said to be completely observable. Otherwise,
the system is said to be unobservable. The necessary and sufficient condition for the
system to be completely observable is that the nxn observability matrix,
Q0  C CF CF2 .. ... CFn 1  T has rank equal to n, ie) rank ( Q 0 ) = n.

4.Outline the block diagram for digital realization of the system described by the
state model

0 1 0 
X (k  1)    X (k )    u (k ) Y (k )  1 1 X ( K )
 2  3  1 

Solution:
X1 ( K  1)  X 2 ( K )
X 2 ( K  1)  2 X1 ( K )  3 X 2 ( K )  u(k )
Y (k )  X 1 ( K )  X 2 ( K )

5.Find state model for the system with difference equation,

y(k  2) 1.5 y(k  1)  0.5 y(k )  r (k  1)  0.5 r (k )

Taking Z transform on both sides,

Y ( z )  Z 2  1.5Z  0.5  R(Z )  Z  0.5

Y ( z) [Z+0.5]
 2
R( z ) [Z -1.5Z+0.5]

 X 1 (k  1)  0 1   X 1 ( K )  0

 X (k  1)   0.5  1.5       u (k )
 2     X 2  K  1 

 X (k ) 
Y (k )   0.5 1  1   5u (k )
 X 2 (k ) 

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.61

 EI6801- Computer control of Process
6.Outline the state space model of discrete time system

The general form of state model for an nth order linear time invariant discrete time
system is,

X  k  1  F X  K   g u  k   state equation
Y k   C X k   D u k   output equation

where, X  k   n 1 state vector

y  k   p 1 output vector
u  k   m 1 input vector
g  nm input matrix
C  pn output matrix
D  pm transmission matrix; F  n  n system matrix

7.List the methods to determine state transition matrix.

i) Similarity transformation

ii) Cayley Hamilton method

iii) Inverse Z transform method

8. Define state variables.

The state variables of dynamic systems are the smallest set of variables which
determine the state of the dynamic system.

9.Give the formula to find the solution of non homogeneous state equation.

X  k  1  F X  k   g u  k  is the non homogeneous state equation.

At k=0,1 we get
X 1  F X  0   g u  0 
X  2   F X 1  g u 1
Substituting X 1 in X  2
X  2  F  F X  0   g u  0   g u 1  F 2 X  0   Fg u  0   g u 1  (1)
Rewriting equation (1) we get,
2 1
X  2   F 2 X  0    F 2i 1 g u  i 
i 0

In general,
k 1
X  k   F k X  0    F k i 1 g u  i   (2)
i 0

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.62

 EI6801- Computer control of Process
10. A discrete time system is characterized by the system matrix

0 1
F  
 2 3
Find the Eigen vector of F and obtain the diagonal representation of ‘F’
Solution:
The characteristic equation of this system is
 1
I  F  0 ; 0
2  3

 2  3  2  0
1  1, 2  2

Eigenvectors V1 & V2 corresponding to the Eigen values 1 & 2 respectively, can be

determined from the ad joint matrix

I  A

  3 1 
Now, adj  I  F   
 2  
2 1 1
For   1  1, adj[1 I  F ]   ; V1   
 2 1  1

1 1 1
  2  2, adj[1I  F ]    ; V   2
 2 2
2
 
The transformation matrix M that transforms ‘F’ into diagonal form is,

1 1
M  
 1 2

The matrix F is,
  2 1   0 1   1 1   2 1  1 1   1 0 
F  M 1FM     2 3  1 2   2 2   1 2   0 2
 1  1       

11. Write the necessary conditions for Jury’s stability test.

Necessary Conditions:
∆z (1) > 0 and
(-1) n ∆z (-1) > 0 (or) ∆z (-1)> 0 for n even
∆z (-1) < 0 for n odd
12. Consider the characteristics polynomial z  z 3  1.3z 2  0.08z  0.24
Check the stability of the system.
∆z(1)= 1-1.3-0.08+0.24=-0.14<0 not satisfied.
Thus the roots don’t lie within the unit circle and the system is unstable.

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.63

 EI6801- Computer control of Process
13.The characteristic equation of z  z 2  z  0.5 . Identify whether the system is
stable or unstable.

∆z(1)= 1-1+0.5=0.5 >0

For ‘n’ even, ∆z(-1)= 1+1+0.5=2.5 >0

As the necessary conditions are satisfied the system is stable.

14.How many rows are formed in Jury’s table and what are the sufficient
conditions to be checked from this table for stability?

Sufficient Condition

For sufficient condition, we construct a table whose elements are based on the
coefficients of ∆z.

∆z = an zn + an-1zn-1 + an-2zn-2 + …….+ a2z2 + a1z + a0 = 0

The table consists of (2n-3) rows, where n is the order of the characteristic equation.

The first column elements of the table are used to check the following (n-1)
conditions. These (n-1) conditions are the sufficient conditions for the stability of the
system.
a0  an
b0  bn1
c0  cn2
…… q0  q2
15. State Schur-Cohn stability criteria.
The characteristic equation for the sampled data system is, 1  A( z )  F ( z)  0
Generally A(z) is the open loop transfer function and will be in the form of a ratio of
two polynomials.
F ( z)  a0  a1 z  a2 z 2  .......  an z n  0
The Schur-Cohn criteria states that all the roots of the characteristic equation lie
inside the unit circle if the following conditions are met.
 k  0 for k odd
k  0 for k even
 k is a determinant that has 2k rows and 2k columns and obtained using the
coefficients of ak .

16. What is a characteristic equation?

The denominator of the closed loop or pulse transfer function i.e.,

1   GhoGH (S )  0 is called the characteristic equation.

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.64

 EI6801- Computer control of Process
17. The characteristic equation of z  z 2  z  0.5 . Examine the stability of the
system using Using Schur-Cohn’s stability test.

The necessary and sufficient conditions that the roots of the characteristic equation lie
inside the unit circle in the Z plane are,

F (0)  1; F (1)  0; F (1)  0

F (0)  0.5  1

F (1)  1 1  0.5  0.5  0

F (1)  1  1  0.5  2.5  0

As the necessary conditions are satisfied the system is stable.

18.For the characteristic polynomial F(z) = 3z4 + 7z3 10z2 + 4z + 1, find the stability
of the system.
3z4 +7z3 +10 z 2 +4z +1= 0

∆z(1)= 3+7+10+4+1=25 >0

∆z(-1)= 1-7+10 - 4+1=1 >0

The Jury table is

row Z0 Z1 Z2 Z3 Z4
1 1 4 10 7 3
2 3 7 10 4 1
3 -8 -17 -20 -5
4 -5 -20 -17 -8
5 39 36 75
a0  an , (i.e) 1  3 satisfied

b0  bn1 , (i.e) 8  5 ; satisfied

c0  cn2 , (i.e) 39  75 ; not satisfied

The system is therefore unstable.

Prepared by Prof.S.Nagammai, HOD/EIE, KLNCE Page 1.65