MATLAB for Control: A Beginners’ Guide

Department of Electrical, Computer & Software Engineering,

The University of Auckland,
Auckland, New Zealand.

MATLAB for Control: A Beginners’ Guide 1 / 66

MATLAB for Control: A Beginners’ Guide 2 / 66
Computation of Inverse Laplace Transform
ilaplace: This function gives the inverse Laplace transform of a function which
is represented by a symbolic expression. The following example illustrate this
Example-1: Determine the inverse Laplace transform of
3s + 1
F (s) =
(s2 + 2s + 9)
]
syms s; % s is a symbolic variable
F s = (3 ∗ s + 1)/(s2 + 2 ∗ s + 9);
f=ilaplace(Fs);
This gives

3 ∗ exp(−t) ∗ (cos(2 ∗ 2(1/2)∗t ) − (2(1/2) ∗ sin(2 ∗ 2(1/2)∗t ))/6)

If we want to determine its value for a particular value of t, (say t = 10) then,
the following command should be executed.
t=10;
y=subs(f);

MATLAB for Control: A Beginners’ Guide 3 / 66

MATLAB for Control: A Beginners’ Guide 4 / 66
Manipulation of Polynomials
How to represent a polynomial ?
I A polynomial is often represented by a vector whose successive elements
correspond to the coefficients of polynomial terms in descending power.
Example-2: The polynomial

s5 + 2s4 + 3s3 + 5s2 + 7s + 9

would be represented by a vector [1.0 2.0 3.0 5.0 7.0 9.0]

However, if we represent this by [9.0 7.0 5.0 3.0 2.0 1.0]

this would correspond to the polynomial

9s5 + 7s4 + 5s3 + 3s2 + 2s + 1

Examples:

x3 + 4x + 1 =⇒ [1.0 0.0 4.0 1.0]

4
10x − 3x =⇒ [10.0 0.0 0.0 -3.0 1.0]

MATLAB for Control: A Beginners’ Guide 5 / 66

MATLAB for Control: A Beginners’ Guide 6 / 66
Manipulation of Polynomials
I A frequently used function in control systems in the conv function.
conv:Convolve two polynomials (represented as row vectors of their coefficients).
It is particularly useful for determining the expanded coefficients for factored
polynomials.
Example: Represent the product of the following two polynomials by a vector

x3 + 3x + 2 x2 + 5x + 6 = f1 (x)f2 (x)

Solution
f1=[1 3 2];
f2=[1 5 6];
f3=conv(f1,f2);
Similarly

f1=[1 9 7 5 2];
x4 + 9x3 + 7x2 + 5x + 2 x2 − 4x + 6 = f1 (x)f2 (x)



=⇒ f2=[1 -4 6];
f3=conv(f1,f2);

MATLAB for Control: A Beginners’ Guide 7 / 66

MATLAB for Control: A Beginners’ Guide 8 / 66
Representation of System by Transfer Functions

tf: This is used to represent a transfer function.

Example-1: Represent the transfer function
s+2
G(s) =
s2 + 5s + 2

G=tf([1 2],[1 5 2]);

Example-2: Represent the transfer function
(s + 3)(s + 5)
G(s) =
s(s + 4)(s + 6)(s2 + 5s + 8)

num=conv([1 3],[1 5]); % Numerator

den1=conv([1 0],[1 4]); % First 2 terms of Denominator
den2=conv([1 6],[1 5 6]); % Last 2 terms of Denominator
den=conv(den1,den2); % Denominator
G=tf(num,den);

MATLAB for Control: A Beginners’ Guide 9 / 66

MATLAB for Control: A Beginners’ Guide 10 / 66
Representation of System by Transfer Functions
poly: This converts roots to polynomial.
poly(V) where V is a vector, is a vector whose elements are the coefficients of
the polynomial whose roots are the elements of V.
Example-1: Consider the function
(x + 1)(x + 2)(x + 3)(x + 5)
I The following command
A=poly([-1 -2 -3 -5]);
would give a vector A=[1 11 41 61 30] which corresponds to the polynomial
x4 + 11x3 + 41x2 + 61x + 30.
Example-2: Represent the transfer function
(s + 3)(s + 5)
G(s) =
s(s + 4)(s + 6)(s + 8)

num=poly([-3 -5]; % Numerator

den=poly([0 -4 -6 -8]); % Denominator
G=tf(num,den);

s2 + 8s + 15
G=
s4 + 18s3 + 104s2 + 192s
MATLAB for Control: A Beginners’ Guide 11 / 66
MATLAB for Control: A Beginners’ Guide 12 / 66
Transfer Function Representation

Given
Y (s) 25
G(s) = = 2
U (s) s + 4s + 25
This can be represented by using
MATLAB Function:tf

Method-I Method-II

num=[0 0 25]; s=tf(’s’)

den=[1 4 25]; G = 25/(s2 + 4 ∗ s + 25);
G=tf(num,den);
Table: Caption

MATLAB for Control: A Beginners’ Guide 13 / 66

MATLAB for Control: A Beginners’ Guide 14 / 66
Transfer Function Representation: Zero Pole Gain Representation

Given
Y (s) 3(s − 1)
G(s) = =
U (s) (s − 2 + i)(s − 2 − i)
This can be represented by using
MATLAB Function:zpk

zeros=[1];
poles=[2-i 2+i];
Gain=3;
G=zpk(zeros,poles,gain);

MATLAB for Control: A Beginners’ Guide 15 / 66

MATLAB for Control: A Beginners’ Guide 16 / 66
Representation of System by Transfer Functions

zpk: This constructs zero-pole model or converts to zero-pole-gain format.

Example-1: Represent the transfer function

s2 + 6s + 5
G(s) =
s4 + 7s3 + 80s2 + 204s
into zero-pole-gain format.

num=[1 6 5]; % Numerator

den=[1 7 80 204 0]; % Denominator
Gtf=tf(num,den); % Transfer function
Gzpk=zpk(Gtf); % zpk is used to create zero-pole-gain model

I This will give

(s + 5)(s + 1)
G(s) =
s(s + 3)(s2 + 4s + 68)

MATLAB for Control: A Beginners’ Guide 17 / 66

MATLAB for Control: A Beginners’ Guide 18 / 66
Representation of System by Transfer Functions

zpk: This constructs zero-pole model or converts to zero-pole-gain format.

Example-2: Represent the transfer function

s4 + 20s3 + 27s2 + 17s + 35

G(s) =
s5 + 8s4 + 9s3 + 20s2 + 29s + 32
into zero-pole-gain format.

num=[1 20 27 17 35]; % Numerator

den=[1 8 9 20 29 32]; % Denominator
Gtf=tf(num,den); % Transfer function
Gzpk=zpk(Gtf); % zpk is used to create zero-pole-gain model

I This will give

(s + 18.59)(s + 1.623)(s2 − 0.214s + 1.16)
G(s) =
(s + 7.042)(s + 1.417)(s2 − 0.4593s + 2.906)

MATLAB for Control: A Beginners’ Guide 19 / 66

MATLAB for Control: A Beginners’ Guide 20 / 66
State Space Representation

Given
   
1 0 1
ẋ = Ax + Bu A= ,B = ,
−2 1 0
   
y = Cx + Du C= 3 −2 , D = 0

MATLAB Function:ss
A=[1 0;-2 1];
B=[1 0]’;
C=[3 -2];
D=[0]; sys=ss(A,B,C,D);

MATLAB for Control: A Beginners’ Guide 21 / 66

MATLAB for Control: A Beginners’ Guide 22 / 66
Partial Fraction Expansion
Given:
2s2 + 3s + 2
G(s) =
s2 + 3s + 2
Represent this function in partial fraction expansion.
MATLAB Function residue
num=[2 3 2];
den=[1 3 2];
[r p k]
=residue(num,den);

I This gives r = [−4 1]0 , p = [−2 − 1]0 , k = 2. This implies

r(1) r(n)
G(s) = k(s) + + ... +
s − p(1) s − p(n)
−4 1
= 2s0 + +
s − (−2) s − (−1)
1 4
=2+ −
s+1 s+2

MATLAB for Control: A Beginners’ Guide 23 / 66

MATLAB for Control: A Beginners’ Guide 24 / 66
Computing Response of Linear Systems

MATLAB for Control: A Beginners’ Guide 25 / 66

Example-1:Simulation of Linear Systems-1
Objective: To compute the output response of the system with the transfer
function
Y (s) s2 + 2s + 3
G(s) = = 3
R(s) s + 2s2 + 3s + 4

for any arbitrary input u(t).

Procedure:
Step-1: Represent the system transfer function

num=[1 2 3]; den=[1 2 3 4];

sys=tf(num,den);
Step-2: Specify the total duration of simulation including the time step of the
integration and input signal of length N number of samples.
N=1000; dt=0.03; % Time step
t=0.0:dt:(N-1)*dt;
t=t’; % Make t a column vector
u=ones(N,1);

MATLAB for Control: A Beginners’ Guide 26 / 66

MATLAB for Control: A Beginners’ Guide 27 / 66
Example-1:Simulation of Linear Systems-2
Step-3: Compute the output response and plot it
y=lsim(sys,u,t);
plot(t,y,’LineWidth’,1.5) grid
xlabel(’Time in Sec’,’FontSize’,18);
ylabel(’Output y(t)’,’FontSize’,18);
title(’Step Response’,’FontSize’,18)
u=ones(N,1);

MATLAB for Control: A Beginners’ Guide 28 / 66

Example-2: Simulation of Linear Systems-1
I Compute both the step and impulse response of the system with the transfer
function
Y (s) s2 + 2s + 3
G(s) = = 3
R(s) s + 2s2 + 3s + 4

Source File: test-lsim2.m

num=[1 2 3]; den=[1 2 3 4];
sys=tf(num,den);N=1000;
t=linspace(0.0,30,N)’;dt=t(2)-t(1);
ustep=ones(N,1);
uimpulse=zeros(N,1);uimpulse(1,1)=1/dt;
ystep=lsim(sys,ustep,t);
yimpulse=lsim(sys,uimpulse,t);
plot(t,ystep,’b’,’LineWidth’,1.5) hold on
plot(t,yimpulse,’r’,’LineWidth’,1.5)
legend(’Step Response’,’Impulse Response’,’FontSize’,18)
grid
xlabel(’Time in Sec’,’FontSize’,18);
ylabel(’Output y(t)’,’FontSize’,18);
title(’Step and Impulse Response’,’FontSize’,18)

MATLAB for Control: A Beginners’ Guide 29 / 66

MATLAB for Control: A Beginners’ Guide 30 / 66
Example-2: Simulation of Linear Systems-2
I The step and impulse response of the system with the transfer function
Y (s) s2 + 2s + 3
G(s) = = 3
R(s) s + 2s2 + 3s + 4
is shown below

MATLAB for Control: A Beginners’ Guide 31 / 66

MATLAB for Control: A Beginners’ Guide 32 / 66
Representation of Cascaded Systems

MATLAB for Control: A Beginners’ Guide 33 / 66

MATLAB for Control: A Beginners’ Guide 34 / 66
Cascaded Systems: Blocks in Series
I Consider the figure below which consists of 2-systems with transfer functions
G1 (s) and G2 (s)

I The relation between the output C(s) and input R(s) is given by
C(s)
= G1 (s)G2 (s) =⇒ C(s) = G1 (s)G2 (s)R(s)
R(s)
I Determine the equivalent block using MATLAB if
s+4 4
G1 (s) = , G2 (s) = 2
s2 + 2s + 10 s + 5s + 2

num1=[1 4]; den1=[1 2 10]; %First Block

num2=[4]; den2=[1 5 2]; %Second Block
sysG1=tf(num1,den1); % First Transfer Function G1 (s)
sysG2=tf(num2,den2); % Second Transfer Function G2 (s)
sysEq=series(sysG1,sysG2); % Equivalent Single Transfer Function

MATLAB for Control: A Beginners’ Guide 35 / 66

MATLAB for Control: A Beginners’ Guide 36 / 66
Cascaded Systems: Blocks in Parallel
I Consider the figure below which consists of 2-systems with transfer functions
G1 (s) and G2 (s)

I The relation between the output C(s) and input R(s) is given by
C(s)
= G1 (s) + G2 (s) =⇒ C(s) = [G1 (s) + G2 (s)] R(s)
R(s)
I Determine the equivalent block using MATLAB if
s+4 4
G1 (s) = 2 , G2 (s) = 2
s + 2s + 10 s + 5s + 2
num1=[1 4]; den1=[1 2 10]; %First Block
num2=[4]; den2=[1 5 2]; %Second Block
sysG1=tf(num1,den1); % First Transfer Function G1 (s)
sysG2=tf(num2,den2); % Second Transfer Function G2 (s)
sysEq=parallel(sysG1,sysG2); % Equivalent Single Transfer Function
MATLAB for Control: A Beginners’ Guide 37 / 66
MATLAB for Control: A Beginners’ Guide 38 / 66
Example-1:Simulating Cascaded Feedback Systems-1

I Consider the figure below which consists of 2-systems with transfer functions
G1 (s) and G2 (s)

I The relation between the output C(s) and input R(s) is given by
 
C(s) G1 (s) G1 (s)
= =⇒ C(s) = R(s)
R(s) 1 + G1 (s)G2 (s) 1 + G1 (s)G2 (s)

Objective: To compute the output c(t) of the closed loop system for step
reference i.e when r(t) = 1(t).

MATLAB for Control: A Beginners’ Guide 39 / 66

MATLAB for Control: A Beginners’ Guide 40 / 66
Example-1:Simulating Cascaded Feedback Systems-2
Step-1: Determine the equivalent block using MATLAB. Let
s+4 4
G1 (s) = 2 , G2 (s) = 2
s + 2s + 10 s + 5s + 2
num1=[1 4]; den1=[1 2 10]; %Feed forward Block
num2=[4]; den2=[1 5 2]; %Feedback Block
sysG1=tf(num1,den1); % First Transfer Function G1 (s)
sysG2=tf(num2,den2); % Second Transfer Function G2 (s)
sysEq=feedback(sysG1,sysG2); % Equivalent Single Transfer Function
Step-2: Simulation of Closed Loop System

Source File: test-lsim-feedback1.m

timestep=0.01;% Integration Timestep
N=1000;
t=0.0:timestep:(N-1)*timestep;t=t’;
r=ones(N,1);
y=lsim(sysEq,r,t);
plot(t,y,’LineWidth’,1.5) grid
xlabel(’Time in Sec’,’FontSize’,16);
ylabel(’Output y(t)’,’FontSize’,16);
title(’Step Response’,’FontSize’,16)
MATLAB for Control: A Beginners’ Guide 41 / 66
MATLAB for Control: A Beginners’ Guide 42 / 66
Example-2:Simulating Cascaded Feedback Systems-1

I Consider the unity feedback system consisting of the controller Gc (s) and system
G(s) which are given by

Objective: To compute the output y(t) of the closed loop system for step
reference i.e when r(t) = 1(t).
Step-1: Determine the equivalent block using MATLAB. Let
s+1 1
Gc (s) = , G2 (s) =
s+2 500s2

numGc=[1 1]; denGc=[1 2]; Gc=tf(numGc,denGc); % Controller Block

numG=[1]; denG=[500 0 0]; G=tf(numG,denG); % System Block
Gseries=series(Gc,G); % Equivalent block in feedforward path
Gclosed=feedback(Gseries,1); % Equivalent Transfer Function

MATLAB for Control: A Beginners’ Guide 43 / 66

MATLAB for Control: A Beginners’ Guide 44 / 66
Example-2:Simulating Cascaded Feedback Systems-2
Step-2: Simulation of Closed Loop System
Source File: test-lsim-feedback2.m
timestep=0.2;% Integration Timestep
N=100000;
t=0.0:timestep:(N-1)*timestep;t=t’;
r=ones(N,1);
y=lsim(Gclosed,r,t);
plot(t,y,’LineWidth’,1.5) grid
xlabel(’Time in Sec’,’FontSize’,16);
ylabel(’Output y(t)’,’FontSize’,16);
title(’Step Response’,’FontSize’,16)

MATLAB for Control: A Beginners’ Guide 45 / 66

MATLAB for Control: A Beginners’ Guide 46 / 66
Concept of Dominant Poles and Zeros

Poles which are closer to the jω-axis are called dominant poles and those
far away from the jω-axis are called non-dominant or insignificant poles

The distance D between the dominant reason and the non-dominant (

insignificant) region is shown in Fig 1. Now comes the important question: how
far from the imaginary axis the poles should be located so that it can be
designated as insignificant or non-dominant. We do not have any clear-cut rules
to determine this. However, as a rule of thumb,

MATLAB for Control: A Beginners’ Guide 47 / 66

MATLAB for Control: A Beginners’ Guide 48 / 66
Dominant Poles and Zeros

Rule of Thumb

If the magnitude of the real part of the pole is at least 5 to 10 times that of
a dominant pole or a pair of complex dominant poles, then the pole may be
regarded as non-dominant (insignificant) from the perspective of preserving
transient response with acceptable error.

a. Similar to the poles, the zeros which are close to the jω-axis in the left-half
s-plane affect the transient response significantly compared to the zeros which
are located far away from the jω-axis ( relative to dominant poles).
b. Although,the regions shown in Figure 1 give us a rough idea about the dominant
and non-dominant regions in the s-plane, while designing the controllers ( e.g.
the pole placement controller), we often desire the dominant poles to be located
near the regions shown in Figure 2 which will ensure ξ = 0.707.

MATLAB for Control: A Beginners’ Guide 49 / 66

MATLAB for Control: A Beginners’ Guide 50 / 66
Dominant Poles & Zeros

Figure: Regions of dominant and insignificant poles in the s-plane for purpose of design

MATLAB for Control: A Beginners’ Guide 51 / 66

MATLAB for Control: A Beginners’ Guide 52 / 66
Correct Procedure of Neglecting the Insignificant Poles
While approximating a higher order system to lower order one, we must ensure
that both the transient and the steady state responses of the original system are
preserved ( with practically acceptable error). The procedure to achieve this are
as follows:
a. Suppose the system is represented as:
K(s + z1 )(s + z2 ) . . . (s + zm )
G(s) = (1)
sN (s + p1 )(s + p2 ) . . . (s + pα )(s + pα+1 ) . . . (s + pn )
b. Assume that there are α-number of insignificant poles located at
s = −p1 , −p2 , . . . , −pα .
c. Represent the terms associated with these poles in time constant form. The
transfer function of Eq. (1) becomes
K (s + z1 )(s + z2 ) . . . (s + zm )
 
G(s) = (2)
p1 p2 . . . pα sN (1 + T1 s)(1 + T2 s) . . . (1 + Tα s)(s + pα+1 ) . . . (s + pn )
1
where Ti = , i = 1, 2, . . . , α.
pi
d. Eliminate the terms (1 + Ti s), i = 1, 2, . . . , α. This gives the approximated
transfer function as:
 
K (s + z1 )(s + z2 ) . . . (s + zm )
G(s) = (3)
p1 p2 . . . pα sN (s + pα+1 ) . . . (s + pn )
MATLAB for Control: A Beginners’ Guide 53 / 66
MATLAB for Control: A Beginners’ Guide 54 / 66
Approximating Higher Order Systems
Consider a system with closed loop transfer function
16000(s + 1)
T(s) = (4)
(s + 2)(s + 8)(s + 40)(s + 50)
The poles at s = −40 and s = −50 are far to the left of jω-axis.Their
contribution to the transient response is negligible ( they decay very quickly to
zero as e−40t and e−50t ). These poles can be neglected without appreciable error
as follows:
We will represent the system as:
16000(s + 1)
T(s) = (5)
1 1
40 × 50(1 + s)(1 + s)(s + 2)(s + 8)
40 50
The approximated transfer function equals to
8(s + 1)
T(s) = (6)
(s + 2)(s + 8)
Note that, following the above method, the steady state value of the original
system is equal to the steady state value of the approximated system. For a unit
step input, it is equal to 0.5. However, if we simply remove the terms (s + 40)
and (s + 50) in Eq. (4), the approximated second order system will have a
different steady state value i.e 1000 to a step input.
MATLAB for Control: A Beginners’ Guide 55 / 66
MATLAB for Control: A Beginners’ Guide 56 / 66
Adding Poles to Closed Loop System:
Original System:
C(s) K K
T(s) = = 2 =⇒ C(s) = R(s)
R(s) s + a1 s + a2 s2 + a1 s + a2
Hence for step input where R(s) = 1/s
K K
c(∞) = css = lim sC(s) = lim = = T(0)
s→0 s→0 s2 + a1 s + a2 a2
Addition of a Pole:
If we add a pole at s = −b to the closed-loop system, the system becomes
C(s) K K
T(s) = = 2 =⇒ C(s) = R(s)
R(s) (s + a1 s + a2 )(s + b) (s2 + a1 s + a2 )(s + b)
Hence for step input where R(s) = 1/s
K K
c(∞) = css = lim sC(s) = lim =
s→0 s→0 (s2 + a1 s + a2 )(s + b) a2 b
Thus the the steady state error of the modified system is reduced. If we want to
have the same steady state error, the gain K should be multiplied by the factor
b.
MATLAB for Control: A Beginners’ Guide 57 / 66
MATLAB for Control: A Beginners’ Guide 58 / 66
Adding Zeros to Closed Loop System:

Addition of a Zero:
If we add a zero at s = −c to the closed-loop system, the system becomes
C(s) K(s + c) K(s + c)
T(s) = = 2 =⇒ C(s) = R(s)
R(s) (s + a1 s + a2 ) (s2 + a1 s + a2 )

Hence for step input where R(s) = 1/s

K(s + c) Kc
c(∞) = css = lim sC(s) = lim =
s→0 s→0 (s2 + a1 s + a2 ) a2
Thus the the steady state error of the modified system is increased. If we want
to have the same steady state error, as of the original system, the gain K of the
modified system should be divided by the factor c.

MATLAB for Control: A Beginners’ Guide 59 / 66

MATLAB for Control: A Beginners’ Guide 60 / 66
MATLAB for Control: A Beginners’ Guide 61 / 66
MATLAB for Control: A Beginners’ Guide 62 / 66
MATLAB for Control: A Beginners’ Guide 63 / 66
MATLAB for Control: A Beginners’ Guide 64 / 66
MATLAB for Control: A Beginners’ Guide 65 / 66
MATLAB for Control: A Beginners’ Guide 66 / 66