Linear Control Systems

Realization theory and Algorithms

– p. 1/40
In summary, a controller-form realization of a multivariable
transfer function H(s) can be determined as follows:
Let dj (s) be the monic least common multiple of all
denominators in the j th column of H(s), and µj be the
degree of dj (s)
Form an m × m diagonal matrix D(s) with dj (s) as the
j th diagonal element
 
1
 s 
Write dj (s) as dj (s) = sµj + Lj 
 
.
. 
 . 
sµj −1

– p. 21/40
  
L1
 L2 
Let L = 
 
 ... 

Lm
Let N (s) = H(s)D(s)
Write the j th column of [N (s) − H(∞)D(s)] as
 
1
 s 
Mj 
 
 .
.. 

sµj −1
h i
Let M = M1 M2 · · · Mm

– p. 22/40
 Let
  

 1 

...

 

  
Āc = Block diag  , i = 1, . . . , m
 




 1  


 0 ··· ··· 0
 

µi ×µi
  

 0 

.. 

 

. 
 
B̄c = Block diag  , i = 1, . . . , m





 0  


1 µ

 

i ×1

– p. 23/40
 A controller-form realization of H(s) is given by
{Ac , Bc , Cc , Dc }, where

Ac = Āc − B̄c L
Bc = B̄c
Cc = M
Dc = H(∞)

Ac and Bc are block diagonal matrices

– p. 24/40
Example:
h 2
i
4s +7s+3 1
H(s) = s3 +4s2 +5s+2 s+1

d1 (s) = s3 + 4s2 + 5s + 2, d2 (s) = s + 1

µ1 = 3, µ2 = 1
" #
(s3 + 4s2 + 5s + 2) 0
D(s) =
0 (s + 1)
 
h i 1 h i
d1 (s) = s3 + 2 5 4  s  ⇒ L1 = 2 5 4
 
s2

d2 (s) = s + 1 ⇒ L2 = 1
– p. 25/40
 " #
2 5 4 0
L=
0 0 0 1

N (s) = H(s)D(s)
" #
h
4s2 +7s+3 1
i (s3 + 4s2 + 5s + 2) 0
= s3 +4s2 +5s+2 s+1 0 (s + 1)
h i
= (4s2 + 7s + 3) 1

H(∞) = 0 ⇒ N (s) − H(∞)D(s) = N (s)

– p. 26/40
  
h i 1 h i
(4s2 + 7s + 3) = 3 7 4  s  ⇒ M1 = 3 7 4
 
s2
h i h i
M2 = 1 ⇒ M = M1 M2 = 3 7 4 1
   
0 1 0 0 0 0
 0 0 1 0   0 0 
Āc =   , B̄c = 
   
0 0 0 0 1 0

   
0 0 0 0 0 1

– p. 27/40
  
0 1 0 0
0 0 1 0
 
Ac = Āc − B̄c L = 
 
−2 −5 −4 0

 
0 0 0 −1
 
0 0
 0 0

Bc = B̄c = 
 
 1 0


0 1
h i h i
Cc = M = 3 7 4 1 , Dc = 0 0

– p. 28/40
Example:
" #
s+1 1
H(s) = s+2 s+1
1 1
s+1 s+1

d1 (s) = (s + 1)(s + 2) = s2 + 3s + 2, d2 (s) = s + 1

µ1 = 2, µ2 = 1
" #
(s2 + 3s + 2) 0
D(s) =
0 (s + 1)
" #
h i 1 h i
d1 (s) = s2 + 2 3 ⇒ L1 = 2 3
s

d2 (s) = s + 1 ⇒ L2 = 1

– p. 29/40
 " #
2 3 0
L=
0 0 1

N (s) = H(s)D(s)
" #" #
s+1 1
s+2 s+1 (s + 1)(s + 2) 0
= 1 1
s+1 s+1 0 (s + 1)
" #
(s2 + 2s + 1) 1
=
(s + 2) 1
" # " #
1 0 −(s + 1) 1
H(∞) = ⇒ N (s)−H(∞)D(s) =
0 0 (s + 2) 1

– p. 30/40
" # " #" # " #
−(s + 1) −1 −1 1 −1 −1
= ⇒ M1 =
(s + 2) 2 1 s 2 1

" # " #
1 h −1 −1 1
i
M2 = ⇒ M = M1 M2 =
1 2 1 1
   
0 1 0 0 0
Āc =  0 0 0  , B̄c =  1 0 
   
0 0 0 0 1

– p. 31/40
  
0 1 0
Ac = Āc − B̄c L =  −2 −3 0 
 
0 0 −1
 
0 0
Bc = B̄c =  1 0 
 
0 1
" # " #
−1 −1 1 1 0
Cc = M = , Dc = H(∞) =
2 1 1 0 0

– p. 32/40
Using duality, we can obtain an observer-form realization of
any proper rational transfer function

Procedure:
Given H(s), let H̃(s) = H T (s)
Find a controller-form realization {Ãc , B̃c , C̃c , D̃c } of
H̃(s)
Let Ao = ÃTc , Bo = C̃cT , Co = B̃cT , Do = D̃cT
{Ao , Bo , Co , Do } is an observer-form realization of
H(s)

– p. 33/40
This is so because

H̃(s) = C̃c (sI − Ãc )−1 B̃c + D̃c

H(s) = H̃ T (s)
 −1
T
= C̃c (sI − Ãc ) B̃c + D̃c
= B̃cT (sI − ÃTc )−1 C̃cT + D̃cT
= Co (sI − Ao )−1 Bo + Do

– p. 34/40
Example:
" # " #
s+1 1 s+1 1
H(s) = s+2
1
s+1
1 , H̃(s) = H T (s) = s+2
1
s+1
1
s+1 s+1 s+1 s+1
   
0 1 0 0 0
Ãc =  −2 −3 0  , B̃c =  1 0 
   
0 0 −1 0 1
" # " #
−1 −1 1 1 0
C̃c = , D̃c =
2 1 1 0 0

– p. 35/40
    
0 −2 0 −1 2
Ao =  1 −3 0  , Bo =  −1 1 
   
0 0 −1 1 1
" # " #
0 1 0 1 0
Co = , Do =
0 0 1 0 0

– p. 36/40
Example:
h 2
i
4s +7s+3 1
H(s) = s3 +4s2 +5s+2 s+1
" 2
#
4s +7s+3
H̃(s) = H T (s) = s3 +4s2 +5s+2
1
s+1

D(s) = d1 (s) = (s3 + 4s2 + 5s + 2)(s + 1)

= s4 + 5s3 + 9s2 + 7s + 2
 
1
h i s 
= s4 + 2 7 9 5  2
 

 s 
s3

– p. 37/40
 h i
L = L1 = 2 7 9 5

N (s) = H̃(s)D(s)
" #
(4s2 + 7s + 3)(s + 1)
=
(s3 + 4s2 + 5s + 2)
 
" # 1
3 10 11 4  s 
 
N (s) − H̃(∞)D(s) =  2 
2 5 4 1  s 
s3
" #
3 10 11 4
M = M1 =
2 5 4 1

– p. 38/40
    
0 1 0 0 0
0 0 1 0 0
   
Āc =   , B̄c = 
   
0 0 0 1 0

   
0 0 0 0 1
   
0 1 0 0 0
 0 0 1 0
  0 
Ãc =  , B̃ =
   
c
 0 0 0 1   0
  

−2 −7 −9 −5 1
" # " #
3 10 11 4 0
C̃c = , D̃c =
2 5 4 1 0

– p. 39/40
    
0 0 0 −2 3 2
1 0 0 −7 10 5
   
Ao =   , Bo = 
   
0 1 0 −9 11 4

   
0 0 1 −5 4 1
h i h i
Co = 0 0 0 1 , Do = 0 0

– p. 40/40