Chapter 12

Design via State Space

 Controller Design
An nth order feedback control system has the following characteristic equation
s n  an 1s n 1  .......  a1s  a0  0
Pole Placement
A plant shown in (a) is
represented in state space by

x  Ax +B u
•

y  Cx
The state equation for
the closed loop system in
(b) can be written as

x  Ax +B u  Ax +B( -Kx  r ) ( A -BK) x +B r

•

y  Cx
a. State-space representation of a plant;
b. plant with state-feedback
 Controller Design

x  Ax +B u
•

y  Cx

x  Ax +B u  Ax +B( -Kx  r ) ( A -BK) x +B r

•

y  Cx

a. Phase-variable representation for plant;

b. plant with state-variable feedback
State-space forms for

C(s)/R(s) =(s+ 3)/[(s+ 4)(s+ 6)].

Note: y = c(t)
 Pole Placement for Plants in Phase-Variable Form
v Represent the plant in phase-variable from.
v Feed back each phase variable to the input of the plant through a gain ki
v Find the charac. Equation for the closed-loop system represented in 2.
v Decide upon all closed-loop pole locations and determine an equivalent
charac. Equation.
v Equate like coefficients of the charac. Equations from 3 & 4 and solve for ki.
Following these steps, the phase-variable representation for the plant is

 0 1 0 .... 0  0 
 0 0 1 .... 0  0 
  
A   ; B  0 ;C  c1 c 2 ... c n 
   
   
 a0 a1 a2 ... an 1  1 

The charac. Equation is s n  an 1s n 1  .......  a1s  a0  0

Pole Placement for Plants in Phase-Variable Form

The system matrix for the closed-loop system is

 0 1 0 .... 0 
 0 0 1 .... 0 
 
A  BK   
 
 
 (a0  k 1 ) (a1  k 2 ) (a2  k 3) ... (an 1  k n ) 
And the charac. Equation by inspection is
det(sI  (A  BK ))  s n  (an 1  k n )s n 1  (an 2  k n 1 )s n 2
+.......  (a1  k 2 )s  (a0  k 1 )  0

Assume the desired charac. Equation for proper pole placement is

s n  d n 1s n 1  d n 2 s n 2  .......  d 2 s 2  d 1s  d 0  0
The feedback coefficients can be found as
k i 1  d i  ai
 Controller design for phase-variable form Example 12.1
Problem Given the plant
20(s  5)
G (s ) 
s (s  1)(s  4)
Design the phase-variable
feedback gains to yield 9.5%
overshoot and settling time of
0.74 second.
Solution: Begin by calculating
the desired closed-loop
charac. Equation, the closed –
loop poles are -5.4+/-j7.2. we
select the third pole at -5.1
close to the zero.

a. Phase-variable representation for plant of Example 12.1;

b. plant with state-variable feedback
Controller design for phase-variable form Example 12.1

The closed-loop system’s state equation from previous figure is

 0 1 0  0 

x   0 0 1  x  0  r ;
   y  100 20 0 x
 k 1 (4  k 2 ) (5  k 3 )  1 
The closed loop system matrix is
 0 1 0 
A  BK   0 0 1 

The closed loop charac. Equation is  k 1 (4  k 2 ) (5  k 3 ) 
det(sI  (A  BK ))  s 3  (5  k 3 )s 2  (4  k 2 )s  k 1  0

This equation must match the desired charac. Equation

s 3  15.9s 2  136.08s  413.1  0

We obtain: k1=413.1, k2=132.08, k3=10.9
 Controller design for phase-variable form Example 12.1

Substitute the values we obtain:

 0 1 0  0 

x   0 0 1  x  0 r ; y  100 20 0 x
 413.1 136.08 15.9 1 

The transfer function is

20(s  5)
T (s ) 
s 3  15.9s 2  136.08s  413.1

The simulation shows 11.5%

overshoot and a settling time of
0.8.
A redesign with 3rd pole
canceling the zero at -5 will yield
performance equal to the
requirements.
 Controllability

If an input to a system can be found that takes every state

variable from desired initial state to a desired final state, the
system is said to be controllable; otherwise, the system is
uncontrollable.
Pole placement is viable design technique only for systems that
are controllable
Controllability by inspection A system with distinct
eignvalues and diagonal system matrix is controllable if the
input coupling matrix B does not have any rows that are zero.

 a1 0 0  1

For figure (a) x   0 a2 1  x  1 u

 0 0 a3  1

 a4 0 0  0 

x   0 a5 1  x  1  u
For figure (b) 
 0 0 a6  1  a. controllable and
b. uncontrollable systems
 Controllability matrix
•
An nth-order plant whose state equation is x  Ax  Bu
Is completely controllable if the matrix C M   B AB A 2 B ... A n 1B 
Is of rank n, where CM is called controllability matrix.
Problem given the system in figure, determine its controllability
Solution: the state equation for the system is  1 1 0  0 

x   0 1 0  x  1  u
The controllability matrix is 0 1 2  0 0 2 1 
C M   B AB A 2 B   1 1 1 
1 2 4 

The rank of CM is 3, so
The system is controllable
 Controller design by matching coefficients

Problem Given a plant Y(s)/U(s)= 10/[(s + 1)(s + 2)], design state feedback to
yield a 15% overshoot with settling time of 0.5 second.
Solution: writing the state equation
for (b) we have.
  2 1  0 
x   x    r; y= 10 0 x
 k 1 (k 2  1)  1 

The characteristic equation is

s 2  (5  k 2 )s  (2  2k 2  k 1 )  0

Compare it to the transient response

requirements s 2  16s  239.5  0

We get k2=13 and k1=211.5

a. Signal-flow graph in cascade form for G(s ) =10/[(s + 1)(s + 2)];

b. system with state feedback added
 Controller design by transformation
Assume a plant not represented in phase-variable form
•
z  A z  Bu
y  Cz
Whose controllability matrix is C Mz   B AB A 2B ... A n 1B 

Assume we can transform it to phase-variable by z =Px so that

•
x  P 1 A Px  P 1 Bu
y  CPx
And controllability matrix C Mx  P 1  B AB A 2B ... A n 1B 

1
We solve for P as
P  C Mz C Mx
 Controller design by transformation Example

Problem design a state-variable feedback controller to yield a 20.8% overshoot

and a settling time of 4 seconds for the plant (s  4)
G (s ) 
(s  5)(s  1)(s  2)

Solution first find state equation and controllability matrix

 5 1 0 0 
z  A z z  B z u   0 1  z   0  u
•
2
 0 0 1 1 
y  C z z   4 1 0 z 0 0 1  Since the det of the matrix

C Mz  Bz Az Bz 
Az2 Bz  0 1  3
is -1 the system is
controllable
1  1 1 
 Controller design by transformation Example
Now we convert the system to phase-variable.
The charac. Equation is det(sI  A z )  s 3  8s 2  17s  10  0
Using the coefficients we write the phase variable representation as
 0 1 0 0
x  A x x  B x u   0 1  x   0  u
•
0
 10 17 8 1 
y   4 1 0 x
0 0 1 
A x B x A x 2 B x   0 1 8
The controllability matrix is
C M x   B x
1 8 47 
Now we can calculate P as  1 0 0
1
P  C Mz C Mx   5 1 0
10 7 1 

For the desired design (20.8% overshoot and settling time of 4 seconds),
the dominant poles would give s 2  2s  5 we choose the third pole at
-4 to cancel the zero. So the total charac. Equation is
D (s )  s 3  6s 2  13s  20  0
 Controller design by transformation Example

The state equation for the phase-variable form with feedback are
 0 1 0 
x  (A x  B x K x )x   x
•
0 0 1 
 (10  k 1x ) (17  k 2 x ) (8  k 3x ) 
y   4 1 0 x

The chara. Equation is

det(sI  (A x  B x K x ))  s 3  (8  k 3x )s 2  (17  k 2 x )s  (10  k 1x )  0
Comparing with desired chara. Equation and solving for Kx
K x   k 1x k 2x k 3x   10 4 2
We can transform the controller back to original system as

K z  K x P 1  20 10 2
The final closed-loop system with state-variable feedback is shown in figure in
next slide
 Designed system with state-variable feedback for Example

To verify our
design, the state
equation for
shown system is
given below and
the closed-loop
transfer function is
obtained as shown
below as well.

 5 1 0 0 
z  ( A z  B z K z ) z  B z r   0 1  x   0  r
•
2
 20 10 1  1 
y  C z z   4 1 0 z

(s  4) 1 So design requirements
T (s )  
s 3  6s 2  13s  20 s 2  2s  5 have been met
Observer Design
If the sate variables are not available because of system configuration
or cost, it is possible to estimate the states. Estimated states are then
fed to the controller. An observer, sometimes called an estimator, is
used to calculate state variables that are not accessible from the plant.
Here the observer is a model of the plant.
Assume a plant And

an observer
 

x  Ax  Bu x  A x  Bu
 
y  Cx y C x
Subtract the observer from the plant, we obtain

  
x  x  A (x  x )
 
y  y  C (x  x )
 Observer Design

State-feedback design
using an observer to
estimate unavailable
state variables:
a. open-loop observer;
b. closed-loop observer;
c. exploded view of a
closed-loop observer,
showing feedback
arrangement to
reduce state-variable
estimation error
 Observer Design
Writing the state equations
of

the observer from figure
  
x  A x  Bu  L ( y  y )
 
y C x
But the state equation for
plant is 
x  Ax  Bu
y  Cx
Subtract, we get

   
(x  x )  A (x  x )  L ( y  y )
 
y  y  C (x  x )

Where x  x is the error between actual and estimated state vectors
 Observer Design
Substitute

the output equation into the state equation, we get
  
(x  x )  (A  LC )(x  x )
 
y  y  C (x  x )
  
Let ex  x  x we have, e  (A  LC )e and y  y  Ce
x x x
Solving for L, the chara. Equation is det[ I  (A  LC )]  0
For an nth-order plant, the chara. Equation A-LC is
s n  (an 1  l1 )s n 1  (an 2  l 2 )s n 2
+.......  (a1  l n 1 )s  (a0  l n )  0
This can be written by inspection if compared to chara. Equation for plant, if
the plant is represented in observer canonical form.
det(sI  A )  s n  an 1s n 1  an  2 s n  2  .......  a1s  a0  0
We equate this to desired chara. Equation, and solve for li’s
s n  d n 1s n 1  d n 2 s n 2  .......  d 2 s 2  d 1s  d 0  0
Third-order observer in observer canonical form:

a. before the addition of feedback;

b. after the addition of feedback
 Observer design for observer canonical form

Problem: Design an observer for the

plant (s  4)
G (s ) 
(s  2)(s  1)(s  5)
s 4
 3
s  8s 2  17s  10
Which is represented in observer
canonical form, the observer will
respond 10 times faster than the
controller loop designed in previous
example.
Solution: the canonical form
representation is first shown (a). Now
form the difference between actual
plant and observer outputs, and add
the feedback to the derivative of each
state variable as shown in (b).
 Observer design for observer canonical form

Next find the chara. Polynomial. The state equation for the estimated plant is
  8 1 0  0
  
x  A x  Bu   17 0 1  x  1  u
 10 0 0   4  and the observer error is
    (8  l1 ) 1 0
y  C x  1 0 0 x 
e x  (A  LC )e x   (17  l 2 ) 0 1  e x
 (10  l 3 ) 0 0
The chara. Polynomial is
det[ I  (A  LC )]  s 3  (8  l1 )s 2  (17  l 2 )s  (10  l 3 )  0
The closed-loop controlled system has dominant 2nd order poles at 1  j 2
To make the observer 10 times faster, we design the observer poles at 10  j 20
We select the 3rd pole at -100, the desired chara. Is then
(s  100)(s 2  20s  500)  s 3  120s 2  2500s  50000

Compare and solve, we get l1=112, l2=2483, and l3=49990

Simulation showing response of observer for an input r(t) =100t:

a. closed-loop;
b. open-loop with observer gains disconnected

The response is slower in (b),

since the observer is a copy of the
plant with different initial
conditions
 Observability

If the initial sate vector, x(t0), can be found from

u(t) and y(t) measured over a finite interval of
time from t0, the system is said to be observable;
otherwise it is said to be unobservable.
Observability is the ability to deduce the state
variables from a knowledge of the input, u(t),
and the output, y(t).
By inspection, for diagonalized systems,
(represented in parallel form) with distinct
eigenvalues, if any column of the output matrix
is zero the system is not observable.
An nth order system is completely observable if
the observability matrix  C  is of rank
n.  CA 
OM   
  a. observable and
 n 1 
CA  b. Unobservable systems
 Observability System of Example 12.6

Problem: determine if the system in figure is observable.

Solution: the state and output equations are
0 1 0  0
x  A x  Bu   0 1  x   0  u
•
0
 4 3 2  1 
y   0 5 1 x
The observability matrix

 C   0 5 1
OM   CA    4 3 3 
CA n 1   12 13 9

has determinant = -344, so it is of

full rank equal to 3. The system
is observable.
 UnObservability System of Example 12.7

Problem: determine if the system in figure is observable.

Solution: the state and output equations are
• 0 1  0 
x  A x  Bu    x   u
 5 21l 4  1 
y  5 4 x

The observability matrix

C   5 4 
OM   
  
CA 20 16 

has determinant = 0, so it is not

of full rank. The system is not
observable.
 Observer design by transformation Example
1
Problem design an observer for the plant G (s ) 
(s  2)(s  1)(s  5)
Represented in cascade form. The closed-loop performance of the observer is
governed by the chara. Equation s 3  120s 2  2500s  50000
Solution first represent the system in its original cascade form.
 5 1 0 0 
z  A z z  B z u   0 1  z   0  u
•
2
 0 0 1 1 
y  C z z  1 0 0 z

The observability matrix is  C   1 0 0

OM   CA    5 1 0
CA n 1   25 7 1 
Determinant =1, the system is observable
 Observer design by transformation Example

The chara. Equation for the plant is

det(sI  A )  s 3  8s 2  17s  10  0
We
• can use the coefficients to form the observer canonical form
x  Ax x  B x u
y Cx x
 8 1 0
where A x   17 0 1  ; C x  1 0 0
 10 0 0   C x   1 0 0
O   C A    8 1 0
The observability matrix for this form is M x  x x   
C x A x   47 8 1 
n 1

We now design the observer for the observer canonical form, first
form
 8 1 0  l1   (8  l1 ) 1 0
A x  L x D x   17 0 1   l 2  1 0 0 =  (17  l ) 0 1 
 2 
 10 0 0  l 3   (10  l 3 ) 0 0
 Observer design by transformation Example

The chara. Polynomial is

det[sI  (A x  L x C x )]  s 3  (8  l1 )s 2  (17  l 2 )s  (10  l 3 )

Equating this polynomial to the desired one,
And solve for s 3  120s 2  2500s  50000

112 
L x   2483 
 49990
 1 0 0
P  O Mz 1O Mx   3 1 0
Now transform back to the original representation

And transforming the feedback  1 1 1 

 112 
L z  PLx   2147 
 47619
The final configuration is shown in next slide
Observer design by transformation Example
Figure 12.19
Observer design
step response
simulation:
a. closed-loop
observer;
b. open-loop
observer with
observer gains
disconnected
 Observer design by matching coefficients- Example

Problem design an observer for the phase variable with transient response described by
  0.7 and n  100 for the plant
407(s  0.916) 407s  372.81
G (s )   2
(s  1.27)(s  2.69) s  3.96s  3.42
Solution first represent the system in phase-variable form as shown in figure (a). For the
plant we have  0 1 
A   ; C  372.81 407 
 3.42 3.96 
Calculating the observability matrix, O M  C CA  shows the plant is observable.
T

Next find the characteristic equation, first we have,

 0 1   l1 
A  LC  
    l  372.81 407 
 3.42 3.96   2
 372.81l1 (1  407l1 ) 
 
  (3.42  372.81l 2 )  (3.96  407 l 2 
)
 Observer design by matching coefficients- Example

Now evaluate det[ I  (A  LC )]  0

   372.81l1 (1  407l1 ) 
det[ I  (A  LC )]  det  
 (3.42  372.81l 2 ) (   3.96  407 l 2 
)
  2  (3.96  372.81l1  497l 2 )  (3.42  84.39l1  372.81l 2 )
0
From the problem statement we want   0.7 and n  100

Thus   140  10000  0 comparing coefficients we find l1 and l2 to be -

2

38.397 and 35.506 respectively. Finally we implement the observer as shown

in figure (b) and using
 0 1 
A   ; C  372.81 407  ;
 3.42 3.96 
0  38.397 
B   ; L  
1   35.506 
 Steady-State Error Design via Integral Control

An additional state variable, xN, has been added at the output

of the leftmost integrator. The error is the derivative of this
variable. Now, from Figure,

Writing the state equations from Figure, we have

 Steady-State Error Design via Integral Control
Eqs. can be written as augmented vectors and matrices. Hence,

Substituting Eqs and simplifying, we obtain

Thus, the system type has been increased, and we can use the characteristic
equation to design K and Ke to yield the desired transient response.
 Example: Design of Integral Control

PROBLEM: Consider the plant of the following Eqs.

a. Design a controller without integral control to yield a 10% overshoot and a

settling time of 0.5 second. Evaluate the steady-state error for a unit step input.
b. Repeat the design of (a) using integral control. Evaluate the steady-state error for
a unit step input.
SOLUTION:
a) Using the requirements for settling time and percent
overshoot, we find that the desired characteristic polynomial is
Since the plant is represented in phase-variable form, the
characteristic polynomial for the controlled plant with
state-variable feedback is
 Example: Design of Integral Control

Equating the coefficients of Eqs. , we have

The controlled plant with state-variable

feedback represented in phase-variable
form is

we find that the steady-state error for a step

input is

b) We now represent the integral-controlled plant as follows:

 Example: Design of Integral Control
Now, we find that the transfer function of the plant is G(s) = l/(s2 + 5s + 3). The
desired characteristic polynomial for the closed-loop integral-controlled system is

Since the plant has no zeros, we assume no zeros for the closed-loop system and
augment the desired eqn with a third pole, (s + 100), which has a real part greater
than five times that of the desired dominant second-order poles. The desired third-
order closed-loop system characteristic polynomial is

The characteristic polynomial for the system of is

Matching coefficients, we obtain
 Example: Design of Integral Control
Substituting these values yields this closed-loop integral controlled system:

In order to check our assumption for the zero, we now apply values and find
the closed-loop transfer function to be

Since the transfer function matches our design, we have the desired transient
response. Now let us find the steady-state error for a unit step input.

Thus, the system behaves like a Type 1 system

 Example: Design of Controller and Observer

PROBLEM: Using the simplified block diagram of the plant for the antenna
azimuth position control system shown in Figure, design a controller to yield
a 10% overshoot and a settling time of 1 second. Place the third pole 10 times as
far from the imaginary axis as the second-order dominant pair.
Assume that the state variables of the plant are not accessible and design an
observer to estimate the states. The desired transient response for the observer is a
10% overshoot and a natural frequency 10 times as great as the system response
above. As in the case of the controller, place the third pole 10 times as far from the
imaginary axis as the observer's dominant second-order pair.

SOLUTION: Controller Design: We first design the controller by finding the

desired characteristic equation. A 10% overshoot and a settling time of 1 second yield
ξ = 0.591 and ωn= 6.77. Thus, the characteristic equation for the dominant poles is
s2 + 8s + 45.8 = 0, where the dominant poles are located at -4 ± j5.46. The third
pole will be 10 times as far from the imaginary axis, or at - 40. Hence, the desired
characteristic equation of the closed-loop system is:
 Example: Design of Controller and Observer

Next we find the actual characteristic equation of the closed-loop system. The
first step is to model the closed-loop system in state space and then find its
characteristic equation. From Figure, the transfer function of the plant is

Using phase variables, this

transfer function is converted
to the signal-flow graph
shown in Figure,

and the state equations are

written as follows:
 Example: Design of Controller and Observer

We now pause in our design to evaluate

the controllability of the system. The
controllability matrix, CM, is

The determinant of CM is - 1 ; thus, the system is controllable.

Continuing with the design of the controller, we show the controller's configuration
with the feedback from all state variables in Figure.
 Example: Design of Controller and Observer
We now find the characteristic equation of the system of Figure, the system matrix,
A - BK, is

Thus, the closed-loop system's characteristic equation is

Matching the coefficients, we evaluate the ki’s as follows:

 Example: Design of Controller and Observer
Observer Design: Before designing the observer, we test the system for observability.
Using the A and C matrices, the observability matrix, OM, is

The determinant of OM is 13253. Thus, OM is of rank 3, and the system is observable.

We now proceed to design the observer. Since the order of the system is not high, we
will design the observer directly without first converting to observer canonical form.
We need first to find A - LC.

are used to evaluate A — LC as follows

The characteristic equation for the observer is now evaluated as

 Example: Design of Controller and Observer

From the problem statement, the poles of the observer are to be placed to
yield a 10% overshoot and a natural frequency 10 times that of the system's
dominant pair of poles. Thus, the observer's dominant poles yield
[s2 + (2 x 0.591x 67.7)s + 67.72] = (s2 + 80s + 4583). The real part of the roots of this
polynomial is -40. The third pole is then placed 10 times farther from the imaginary axis
at -400. The composite characteristic equation for the observer is

Matching coefficients, we solve for the observer gains:

Figure shows the completed design, including the controller and the observer.