Modern Control
Tutorial #9
Design of Linear State Feedback Control Laws
Faculty of Engineering and Material Science
GUC Department of Mechatronics Modern Control MCTR 702
Dr. Ayman Ali El-Badawy
Tutorial Contents
• Controller/Compensator Design using state feedback
– Bass Gura
– Ackermann
• Shaping the Dynamic Response
– 1st order systems
– 2nd order systems
– Higher order systems
GUC Faculty of Engineering and Material Science
Department of Mechatronics
Modern Control MCTR 702
Dr. Ayman Ali El-Badawy
Tutorial Contents
• Controller/Compensator Design using state feedback
– Bass Gura
– Ackermann
• Shaping the Dynamic Response
– 1st order systems
– 2nd order systems
– Higher order systems
GUC Faculty of Engineering and Material Science
Department of Mechatronics
Modern Control MCTR 702
Dr. Ayman Ali El-Badawy
State Feedback Control Law
Recall from classical control,
In this feedback system, we can select the gain K to achieve a certain closed
loop system performance. i.e. the gain K can be chosen to place the poles of
the closed loop system, and hence the performance.
GUC Faculty of Engineering and Material Science
Department of Mechatronics
Modern Control MCTR 702
Dr. Ayman Ali El-Badawy
Open loop system
x (t ) A x(t ) B u (t )
y (t ) C x(t ) D u (t )
Closed loop system
By choosing u (t) as
r(t) +
-
x (t ) A BK x(t ) B r (t )
y (t ) C x(t ) D u (t )
GUC Faculty of Engineering and Material Science
Department of Mechatronics
Modern Control MCTR 702
Dr. Ayman Ali El-Badawy
Our goal is to select the gain matrix K so that the new system (A-BK) has the
eigenvalues where we want them rather than using those of the original A - matrix
Hence, Our Task in this tutorial is to find the gain Matrix K
Important Note
To place all the eigenvalues of the closed loop system at the desired positions, the
system should be completely controllable.
The ability to place the poles
controllability
anywhere through state feedback
GUC Faculty of Engineering and Material Science
Department of Mechatronics
Modern Control MCTR 702
Dr. Ayman Ali El-Badawy
Tutorial Contents
• Controller/Compensator Design using state feedback
– Bass Gura
– Ackermann
• Shaping the Dynamic Response
– 1st order systems
– 2nd order systems
– Higher order systems
GUC Faculty of Engineering and Material Science
Department of Mechatronics
Modern Control MCTR 702
Dr. Ayman Ali El-Badawy
1. Bass-Gura Formula:
K 0 a0 1 a1 n 1 an 1 TCCF
1 1
, TCCF P PCCF
where
0 , 1 , 2 ,, n 1 are the desired closed loop characteristic polynomial coefficients
a0 , a1 , a2 ,, an 1 are the open loop characteristic polynomial coefficients
Problem 1:
It is clear that both systems are controllable
GUC Faculty of Engineering and Material Science
Department of Mechatronics
Modern Control MCTR 702
Dr. Ayman Ali El-Badawy
Solution (a): To solve using Bass-Gura formula, we follow the following steps
First, we find the controllability matrix P
1 1
P B AB Controllable system
1 4
Second, we find the closed loop desired characteristic polynomial.
( s) ( s (2 3i)) ( s (2 3i))
s 2 4s 13
Third, we find the characteristic polynomial of the given system, a(s) (det(sI-A))
a( s) sI A
s 2 5s 4
GUC Faculty of Engineering and Material Science
Department of Mechatronics
Modern Control MCTR 702
Dr. Ayman Ali El-Badawy
Fourth, we find the gain matrix for the CCF system
K CCF [ 0 a0 1 a1 ]
[(13 4) (4 5)]
[9 1]
Fifth, we find the gain matrix for the given system by transforming KCCF through.
1 1 1 a1 1 5 1
P P CCF
1 4 1 0 1 0
1 1
4 1
1
TCCF 3 3
4
TCCF
1 1 1
3 3
GUC Faculty of Engineering and Material Science
Department of Mechatronics
Modern Control MCTR 702
Dr. Ayman Ali El-Badawy
Finally, the gain Matrix is then given as
1
K KCCFTCCF 3.3 4.3
To check the answer, we get the eigenvalues of the new A matrix
4.3 4.3
A BK
3.3 0.3
Which has these eigenvalues
2 3i, 2 3i The desired eigenvalues
GUC Faculty of Engineering and Material Science
Department of Mechatronics
Modern Control MCTR 702
Dr. Ayman Ali El-Badawy
Solution (b):
First, we find the controllability matrix P
0 1
P
Controllable system
1 8
Second, we find the closed loop desired characteristic polynomial.
( s ) s 2 9 s 20
Third, we find the characteristic polynomial of the given system (det(sI-A))
a ( s ) s 2 8s 6
GUC Faculty of Engineering and Material Science
Department of Mechatronics
Modern Control MCTR 702
Dr. Ayman Ali El-Badawy
Fourth, we find the gain matrix for the CCF system
K CCF [14 1]
Fifth, we find the gain matrix for the given system by transforming KCCF through.
0 1 1 a1 1 8 1
P P
CCF
1 8 1 0 1 0
1 0 1 1 0
TCCF T CCF
Here, the system is given in
0 1 0 1 the CCF, the KCCF = K
GUC Faculty of Engineering and Material Science
Department of Mechatronics
Modern Control MCTR 702
Dr. Ayman Ali El-Badawy
Finally, the gain Matrix is then given as
KK 1
T
CCF CCF 14 1
To check the answer, we get the eigenvalues of the new A matrix
0 1
A BK
20 9
Which has these eigenvalues
4, 5 The desired eigenvalues
GUC Faculty of Engineering and Material Science
Department of Mechatronics
Modern Control MCTR 702
Dr. Ayman Ali El-Badawy
Tutorial Contents
• Controller/Compensator Design using state feedback
– Bass Gura
– Ackermann
• Shaping the Dynamic Response
– 1st order systems
– 2nd order systems
– Higher order systems
GUC Faculty of Engineering and Material Science
Department of Mechatronics
Modern Control MCTR 702
Dr. Ayman Ali El-Badawy
2. Ackermann’s Formula
GUC Faculty of Engineering and Material Science
Department of Mechatronics
Modern Control MCTR 702
Dr. Ayman Ali El-Badawy
Problem 2:
Repeat problem 1 using Ackermann’s formula
Solution: To solve using Ackermann’s formula, we follow the following steps
First, we find the controllability matrix P
1 1
P B AB
1 4
Second, we find the closed loop desired characteristic polynomial, and hence α’s
( s) ( s (2 3i)) ( s (2 3i))
s 2 4s 13
GUC Faculty of Engineering and Material Science
Department of Mechatronics
Modern Control MCTR 702
Dr. Ayman Ali El-Badawy
Third, by using α’s , we find the matrix α(A)
( A) A2 1 A 0 I
1 0 1 0
2
1 0
4 13
0 4 0 4 0 1
10 0
0 13
Finally, we find the gain matrix using the formula
K [0 1]P 1 ( A)
1
1 1 10 0
[0 1] 3.3 4.3
1 4 0 13
GUC Faculty of Engineering and Material Science
Department of Mechatronics
Modern Control MCTR 702
Dr. Ayman Ali El-Badawy
Solution:
First, we find the controllability matrix P
0 1
P B AB
1 8
Second, we find the closed loop desired characteristic polynomial, and hence α’s
(s) (s (4)) (s (5))
( s ) s 2 9 s 20
GUC Faculty of Engineering and Material Science
Department of Mechatronics
Modern Control MCTR 702
Dr. Ayman Ali El-Badawy
Third, by using α’s , we find the matrix α(A)
( A) A2 1 A 0 I
2
0 1 0 1 1 0
9 20
6 8 6 8 0 1
14 1
6 6
Finally, we find the gain matrix using the formula
K [0 1]P 1 ( A)
1
0 1 14 1
[0 1] 14 1
1 8 6
6
GUC Faculty of Engineering and Material Science
Department of Mechatronics
Modern Control MCTR 702
Dr. Ayman Ali El-Badawy
Tutorial Contents
• Controller/Compensator Design using state feedback
– Bass Gura
– Ackermann
• Shaping the Dynamic Response
– 1st order systems
– 2nd order systems
– Higher order systems
GUC Faculty of Engineering and Material Science
Department of Mechatronics
Modern Control MCTR 702
Dr. Ayman Ali El-Badawy
Shaping the Dynamic Response
1st order system:
•Here, T is called the (time constant).
•So, by choosing the pole of the 1st order
system we can adjust the time constant.
GUC Faculty of Engineering and Material Science
Department of Mechatronics
Modern Control MCTR 702
Dr. Ayman Ali El-Badawy
2nd order systems
Which has the poles
Step response of an underdamped 2nd order system.
Rise time tp Mp 1%
1
0.9
Settling time
0.1
Peak time
tr
ts
Over shoot (maximum peak) M e n d e 1 2
, 0 1
p
So, by playing with the poles, we can play with the response behavior.
GUC Faculty
Science
of Engineering and Material Control Engineering ENME
503
Higher order system:
We can approximate the behavior of high order systems to the behavior of 1st or 2nd
order systems, by placing the dominant poles at the appropriate positions.
Example
Here, the time constant is 0.5 which
leads to a pole at –(1/0.5) = – 2
(one possible solution)
DomPole = -2
Poles2 = -2 -20 For approximate behavior for higher order
systems, we choose the rest of the poles
Poles3 = -2 -20 -21
as away from the imaginary axis as
Poles4 = -2 -20 -21 -22 possible so that their effect is very little.
GUC Faculty of Engineering and Material Science
Department of Mechatronics
Modern Control MCTR 702
Dr. Ayman Ali El-Badawy
Step Response
0.8
Amplitude
0.6
0.4
0.2
0
0 0.5 1 1.5 2 2.5 3 3.5 4
Time (sec)
GUC Faculty of Engineering and Material Science
Department of Mechatronics
Modern Control MCTR 702
Dr. Ayman Ali El-Badawy
/ 1 2 6 Then the poles of the second order system will be
Mp e 0.6671
100
4.6
ts 4 n 1.72
n s1, 2 0.6671*1.72 j1.72 * 1 0.66712
s1, 2 1.14 j1.28
(one possible solution)
DomPoles =
-1.14 + 1.28i
-1.14 - 1.28i
Poles3 = -1.14 + 1.28i -1.14 - 1.28i -10
Poles4 = -1.14 + 1.28i -1.14 - 1.28i -10 -11
GUC Faculty of Engineering and Material Science
Department of Mechatronics
Modern Control MCTR 702
Dr. Ayman Ali El-Badawy
Unit Step response:
0.8
Dom 2nd
0.6 Aug 3rd
ydes
Aug 4th
0.4
0.2
0
0 1 2 3 4 5 6 7 8
time (sec)
GUC Faculty of Engineering and Material Science
Department of Mechatronics
Modern Control MCTR 702
Dr. Ayman Ali El-Badawy
Problem 3:
Consider the single-input, single-output rotational mechanical system
shown. Design a state feedback control law to achieve 0.8 sec settling
time, 3% overshoot.
GUC Faculty of Engineering and Material Science
Department of Mechatronics
Modern Control MCTR 702
Dr. Ayman Ali El-Badawy
Solution:
State Space Model:
GUC Faculty of Engineering and Material Science
Department of Mechatronics
Modern Control MCTR 702
Dr. Ayman Ali El-Badawy
0 1 0
A , B , C 1 0 Open-loop system
40 4 1
Open Loop response:
The open loop unit step response was
obtained numerically using MATLAB as it
will be later shown.
GUC Faculty of Engineering and Material Science
Department of Mechatronics
Modern Control MCTR 702
Dr. Ayman Ali El-Badawy
Controller Design:
Now, we find the desired closed loop poles
/ 1 2 Desired C.L
M p 3% e 0.75 eigenvalues
4.6
Ts 0.8 n 5.75 n 7.67 [-5.71+ 5.12i ; -5.71 - 5.12i]
n
Dsired Characteristic Equation : s 2 2 n s n s 2 11.5s 58.83
2
Then, we find the gain vector K using Ackermann’s formula
0 1
P
1 4
18.83 7.5
( A) A2 11.5 A 58.83I
300 11.17
K 0 1P 1 ( A) 18.86 7.43
GUC Faculty of Engineering and Material Science
Department of Mechatronics
Modern Control MCTR 702
Dr. Ayman Ali El-Badawy
Open-loop system
x (t ) A x(t ) B u (t )
y (t ) C x(t )
0 1 0
A , B , C 1 0
40 4 1
Closed-loop system
0 1 0 0 1
A BK 18.86 7.43
40 4 1 58.86 11.43
0
B , C 1 0
1
GUC Faculty of Engineering and Material Science
Department of Mechatronics
Modern Control MCTR 702
Dr. Ayman Ali El-Badawy
We can find the step response of the open-loop and the closed-loop systems using the
following MATLAB code.
clear, clc
% Define your constants
J=1; b=4; kR=40;
% Define the open loop system
A=[0 1;-kR/J -b/J]; B=[0;1/J]; C=[1 0]; D=[0];
rot_sys = ss(A,B,C,D); % the system model
% Open unit step loop response
t= [0:0.01:3]; %the time vector
u= ones(1,length(t)); %the input vector
x0= [0;0]; %Zero Initial Conditions
[y,t,x]= lsim(rot_sys,u,t,x0);
plot(t,y,'k','linewidth',2);
ylabel('angle (rad)');
hold on
% Controller Design - pole placement
DesEig=[-5.7143 + 5.1195i;-5.7143 - 5.1195i];
K=place(A,B,DesEig) % Compute state feedback gain matrix K3
% we can also use K=acker(A,B,DesEig)
% Closed loop unit step System
Ac= A-B*K; Bc= B; Cc= C; Dc= D; % Compute closed-loop state feedback system
rot_sys_closed = ss(Ac,Bc,Cc,Dc); % Create the closed-loop state-space system
% Closed loop Response
[yc,t,xc] = lsim(rot_sys_closed,u,t,x0);
plot(t,yc,'b','linewidth',2);
legend('open loop','closed loop')
grid on
hold off
And the result will be
0.035
open loop
closed loop
0.03
0.025
0.02
angle (rad)
0.015
0.01
0.005
0
0 0.5 1 1.5 2 2.5 3
It can be easily seen that, when the reference input r = 1, there is steady state error.
i.e. the final value of the output is not equal to 1.
GUC Faculty of Engineering and Material Science
Department of Mechatronics
Modern Control MCTR 702
Dr. Ayman Ali El-Badawy
