You are on page 1of 17

Advanced Control Systems

Lecture 08

Muhammad Mehdi, Ph.D.


Electrical Engineering, BUITEMS
July 12, 2023
Overview

1 State Feedback

2 References

Muhammad Mehdi, Ph.D. 2 / 16


State Feedback

Muhammad Mehdi, Ph.D. 3 / 16


State Feedback
For(an n-dimensional state equation
ẋ = Ax + Bu
y = Cx + Du
The control signals u = r − Kx provide the state feedback control strategy, and the
gain matrix/vector K forces the eigenvalues of the closed-loop to be at the desired
values.

Muhammad Mehdi, Ph.D. State Feedback 4 / 16


State Feedback

Two typical control problems:


▶ The regulator problem, in which r = 0 and aims to keep limt−→∞ y(t) = 0
(Its objective is to maintain the zero output, i.e., a pure stabilization
problem).
▶ The trakcing problem, the closed-loop in which y(t) is specified to track
r ̸= 0.
When r = constant ̸= 0, constant, the regulator and tracking problems are
essentially the same. Tracking a nonconstant reference is a more complex
problem, called the servomechanism problem.

Muhammad Mehdi, Ph.D. State Feedback 5 / 16


State Feedback

Closed-loop

Regulator: Tracking:
Let u = −Kx, then Let u = r − Kx, then
ẋ = (A − BK)x ẋ = (A − BK)x + Br
The solution would be The solution depends on signal r.
x = e(A−BK)t x(0) The stability and transient response
The stability and transient response characteristics depend on the type of
characteristics are determined by the servo system.
eigenvalues of matrix A − BK.

Muhammad Mehdi, Ph.D. State Feedback 6 / 16


State Feedback

Determination of gain matrix/vector


▶ Direct substitution
▶ Transformation method
▶ Ackermann’s formula

Muhammad Mehdi, Ph.D. State Feedback 7 / 16


State Feedback

Direct Substitution (Usually for lower order system (n ≤ 3))

To determine state feedback gain K = [k1 k2 k3 ]


for the desired eigenvalues P = [µ1 µ2 µ3 ]
we need to solve the following equation:

|sI − A + BK| = (s − µ1 )(s − µ2 ) . . . (s − µn )

Muhammad Mehdi, Ph.D. State Feedback 8 / 16


State Feedback

Direct Substitution

E.g. (Regulator problem),


Determine the state feedback gain K for
the desired
 closed-loop poles at
p=  −2 + j4 −2 −j4 −10 when, 
0 1 0 0
A= 0 0 1  , B = 0,
−1 −5 −6 1

Muhammad Mehdi, Ph.D. State Feedback 9 / 16


State Feedback

Direct Substitution
E.g. (Regulator Problem),  
Determining the gain K = k1 k2 k3 . The desired characteristic equation from
the desired closed-loop poles p
(s + 2 − j4)(s + 2 + j4)(s + 10) = s3 + 14s2 + 60s + 200

The closed-loop characteristic equation with the gain is K is


|sI − A + BK| = s3 + (6 + k3 )s2 + (5 + k2 )s + 1 + k1

equating both characteristic equations gives  


6 + k3 = 14, 5 + k2 = 60, 1 + k1 = 200 =⇒ K = 199 55 8

Muhammad Mehdi, Ph.D. State Feedback 10 / 16


State Feedback
Transformation Method
Step 1: Check the controllability, if the system is controllable then,
Step 2: Form the characteristic polynomial of matrix A, i.e.,
|sI − A| = sn + a1 sn−1 + · · · + an−1 s + an
Step 3: Determine the transformation matrix T = M W that transforms the system
state equation into the controllable canonical form. Note: T = I if the system is
already in controllable canonical form.  
an−1 an−2 . . . a1 1
an−2 an−3 . . . 1 0
 
M = [B | AB | . . . | An−1 B], and W =  ... .. .. .. 

 . . . 
 a1 1 . . . 0 0
1 0 ... 0 0

Muhammad Mehdi, Ph.D. State Feedback 11 / 16


State Feedback

Transformation Method
Step 4: Write the desired characteristic polynomials for the desired eigenvalues as:
(s − µ1 )(s − µ2 ) . . . (s − µn ) = sn + α1 sn−1 + · · · + αn−1 s + αn
and determine the values of αi .
Step 5: The required gain matrix K can be determined by:
K = [αn − an | αn−1 − an−1 | | . . . α2 − a2 | α1 − a1 ]

Muhammad Mehdi, Ph.D. State Feedback 12 / 16


State Feedback

Ackermann’s Formula
K = [0 0 . . . 0 1][B | AB | . . . | An−1 B]−1 ϕ(A)

Where ϕ(A) is the system characteristic polynomial.

Muhammad Mehdi, Ph.D. State Feedback 13 / 16


State Feedback
Servo System/Tracking Problem: Type 1

For a system with n−state, the ẋ = (A − BK)x + Bk1 r


state-feedback control input would be
u = −Kx − k1 r
u=  
x1
   x2 

− 0 k2 k3 . . . kn  ..  + k1 (r − x1 )

.
x
  n
K = k1 k2 . . . k n
Closed-loop system would be

Muhammad Mehdi, Ph.D. State Feedback 14 / 16


State Feedback
Servo System/Tracking Problem: Type 1
E.g., (Tracking Problem)

   
0 1 0 0 Matlab implementation:
A = 0 0 1  , B = 0 ,
A = [0 1 0; 0 0 1; 0 -2 -3];
 0 −2  −3 1 1
2 B = [0; 0; 1];
C= 1 0 0 3 p = [ -2+ j *2 , -2 - j *2 , -10];
4 K = acker (A ,B , p )
u = −(k2 x2 + k3 x3 ) + k1 (r − x1 ) 5 K =
6 80 46 11

Muhammad Mehdi, Ph.D. State Feedback 15 / 16


References

Muhammad Mehdi, Ph.D. 16 / 16


References

Muhammad Mehdi, Ph.D. References 16 / 16

You might also like