Chapter 2

Mathematical Modeling of Control

Systems
 2-1 Introduction

To understand and control complex systems, one

must obtain quantitative mathematical models of
these systems.

A mathematical model is defined as a set of

equations that represents the dynamics of the
system accurately, or at least fairly well.
The dynamics of many systems, whether they are
mechanical, electrical, thermal, and so on, may be
described in terms of differential
equations—using physical laws governing a
particular system such as Newton’s laws and
Krichhoff’s laws.
1. Mathematical models
Example. RC Network

Mathematical model: by using the relationship

dy
i C
dt

we have dy
RC y  u
dt
which is a first-order differential equation.
Mathematical models may include many different
forms. For example:

1. Systems described by linear differential

equations;
2. Systems described by nonlinear differential
equations;
3. Systems described by partial differential
equations;
4. …….

Which model is better depends on the particular

system and the particular circumstances.
 2. Linear systems
Example. Spring-mass-damper mechanical system.
Such a system could represent an automobile shock
absorber. We assume the friction force is
proportional to the velocity of the mass and
2
d y (t ) d y (t )
M  r (t )  d  ky ( t )
dt dt
friction k
constant d y (t )
b v (t ) 
y dt
d v (t ) t
M  bv ( t )  k  v ( t )d t  r (t )
r(t) Force dt 0
Rearranging the above differential equation yields
2
d y (t ) d y (t )
M d  ky ( t )  r ( t )
dt dt

It is clear that both sides are the linear functions of

either y, y’ ,y’’ or r. Such a system is called a linear
system.
In classical control theory, a linear system can be
generally described by the following differential
equation ( n ) ( n 1)
a 0y  a 1y    a n 1y  a n y
 b0 x ( m )  b1x ( m 1)    bm 1x  bm x
Then we say the system is an nth order linear
system.
3. Linear time invariant systems and linear time
varying systems

Consider the following linear differential equation

1)
a 0y
(n )
 a 1y (n
   a n 1y  a n y
1)
 b0 x (m )
 b1x (m
   bm 1x  bm x

1) The system is said to be linear time invariant

if all its coefficients are constants;

2) The system is said to be linear time varying if

at least one of its coefficients is the function
of time.
Example. Consider the following two systems:
y
(3)
 2 y ''  3y ' 4 y  r ( t )

and
y
(3)
 si n( t ) y ''  t 2 y ' 4ty  r ( t )

By the above definition with respect to LTI and LTV

systems, it is obvious that the latter is an LTV system.
 2-2 Transfer function and impulse-
response function
1. Integral Transformation

a) Laplace Transformation

Given the real function f(t) with f(t)=0 for t<0. Then
the Laplace transform of f(t) is defined as

L [ f (t ) ] F (s )  f ( t )e
st
dt
0

where s is a complex variable.

Example. Let f(t)=e t, t 0. Find [f(t)].
Example. Step function
A , t  0 A
f (t ) 
0, t  0 t

Find [f(t)].

Example. Ramp function

At
A t , t  0
f (t ) 
0, t  0 t

Find [f(t)].
Example. Sinusoidal function
A si n wt , t  0
f (t ) 
0, t  0

Find [f(t)].
Similarly, for L [ A c o s wt ]

Find [f(t)].

Example. Translated function f(t ):

t t
f(t) f(t )
Let [f(t)]=F(s). Then
[f(t )]=e sF(s), for 0.

Example. Pulse function

A
 , 0  t  t0
f (t )  t 0 A/t0
0, t  0, t  t t
 0 t0

f(t) can be expressed as

A A
f (t )  1( t )  1( t t0 )
t0 t0
A A st 0
L [ f (t ) ]   e
t 0s t 0s
Example. Impulse function
 , t  t 0
d (t t0 ) 
0, t  t 0 t
t0 
t0

 d (t  t
t0 
0
)d t 1

In particular,
0

 d (t )d t  1
0

Then we have
st 0
L [ d (t t0 )]  e
and
L [ d (t ) ] 1
b) Inverse Laplace transformation:
Given the Laplace transform F(s), the operation of
obtaining f(t) is called the inverse Laplace
transformation, and is denoted by
f(t)= 1[F(s)]
The inverse Laplace transform integral is given as

1 c j 
 
st
f (t ) F ( s )e d s
c j 
2p j

where c is a real constant that is greater than the

real parts of all the singularities of F(s).
Remark. Evaluating the inversion integral appears
complicated. In practice, we seldom use this integral
for finding f(t). There are simpler methods for finding
f(t).
Note that these simpler methods for finding inverse
Laplace transforms are based on the fact that the
unique correspondence of a time function and its
inverse Laplace transform holds for any continuous
time function.

Example. Given F(s)=1/(s+1). Find f(t).

Example. Given F(s)=1/s. Find f(t).

c) Laplace Transform Theorems

Theorem 1. Let k be a constant, and F(s) be the

Laplace transform of f(t). Then

L [ kf ( t ) ]  kF (s )

Theorem 2. Let F1(s) and F2(s) be the Laplace

transform of f1(t) and f2(t), respectively. Then

L [ a 1 f1 ( t )  a 2 f 2 ( t ) ]  a 1F 1 ( s )  a 2 F 2 ( s ) , a 1 , a 2  R
Theorem 3. (Differentiation Theorem)

Let F(s) be the Laplace transform of f(t), and f(0) is

the limit of f(t) as t approaches 0. The Laplace
transform of the time derivative of f(t) is

d f ( t ) 
L
 d t   sF (s )  li m
t 0
f (t )  sF (s )  f (0)

In general, for higher-order derivatives of f(t),

d n f ( t )  n  n 1 n 2 d f ( t ) d
n 1
f (t ) 
L  dt n   s F ( s )  li m s
t 0
f (t )  s  n 1 
   dt dt 
n 1 n  2 (1) ( n 1)
snF (s )  s f (0) s f ( 0)  f (0)
Theorem 4. (Integration Theorem)
The Laplace transform of the first integral of f(t)
with respect to t is the Laplace transform of f(t)
divided by s; that is,

 t f ( t )d t   F (s )
 o
L
 s

For nth-order integration,

t t F (s )

  f ( u ) (d u )  n
n
0 0
s
n
 Theorem 5. (Complex Shifting Theorem)

The Laplace transform of f(t) multiplied by e t ,

where is a constant, is equal to the Laplace
transform F(s) with s replaced by s ; that is,

L [e
a t
f (t ) ] F (s a )
Theorem 6. (Real Convolution Theorem)

Let F1(s) and F2(s) be the Laplace transform of f1(t)

and f2(t), respectively, and f1(t)=0, f2(t)=0, for t<0,
then

 L [ f1 ( t )  f 2 ( t ) ]  L  f1 ( t ) f 2 ( t  t )d t 
 t
F 1 (s ) F 2 (s )
 0 
d) Inverse Laplace transformation: partial-
fraction expansion
Partial-fraction expansion
Consider the function
B (s )
F (s ) 
A (s )

where A(s) and B(s) are real polynomials of s. It is

assumed that deg(A(s)) deg(B(s)). Write A(s) as
n 1
A (s )  s  a 1s
n
   a n 1s  a n
whose roots are called poles; write
1
B (s )  b0s  b1s
m m
   bm 1s  bm
whose roots are called zeros.
• Partial-fraction expansion when F(s) involves
distinct poles only. In this case, F(s) can be written
as
B (s ) a1 a2 an
F (s )    
(s  p1 ) (s  p 2 ) (s  p n ) s  p1 s  p2 s  pn
where
ai   (s  p i ) F (s )  s  p i

 p1t  p 2t  pn t
f (t )  a 1e  a 2e    a ne

Example.
s 3
F (s ) 
(s  1) ( s  2 )
Example.
2s12 1 0  2 ( s  1)
F (s )  2 
s  2s  5 ( s  1)  2
2 2

10 2 ( s  1)
 
( s  1)  2 ( s  1)  2
2 2 2 2

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

Therefore,
t t
f (t )  5e si n 2 t  2e c o s 2t , t 0
• Partial-fraction expansion when F(s) involves
Multiple poles. If r of the n roots of A(s) are identical,
G(s) is written
B (s )
G (s )  , (i  1, 2,  , n  r )
(s  p 1 ) ( s  p 2 )  ( s  p n r ) ( s  p i ) r

then G(s) can be expanded as

a1 a2 a n r
G (s )     
s  p1 s  p2 s  p n r
|n r term s o f si m p le ro o ts |
A1 A2 Ar
   
s  pi (s  pi ) 2
(s  pi )r
| r term s o f rep eated ro o ts |
where
aj  ( s  p j )G (s ) 
 s  p j
(j  1, 2,  , n  r )

The determination of the coefficients that correspond

to the multiple-poles is described as follows.

A r  ( s  p i ) r G ( s )  s  p
   i

 1 d
k

A r k  k 
 ( s  p )
r
G (s ) 
 s  p i
(k  1, 2, r  1)

i
k ! ds
 Example. 1
G ( s) 
s( s  1) 3 ( s  2 )

G(s) can be written as

a1 a2 A1 A2 A3
G (s )     
s s 2 s 1 (s  1) 2
(s  1) 3

The coefficients corresponding to the simple roots and

those of the third-order root are

A 3  ( s  1) G ( s )  s 1  1
3

 1 
a 1   sG ( s )  s 0

2

 2
A 
d
 ( s  1)
3
G ( s )  s 1  0
 d s
 
1 
a   ( s  2 )G (s )    1 d
2
 2 s 2
2  1
A   ( s  1)
3
G ( s )  s 1  1
2  
 2 ds
 e) Solving LTI differential equations

• Transform the differential equation to the s-

domain by Laplace transform.
• Manipulate the transformed algebraic equation
and solve for the output variable.
• Perform partial-fraction expansion to the
transformed algebraic equation.
• Obtain the inverse Laplace transform.
Example.
y  3y  2 y  5u ( t )
where
u (t )  1( t ) , y ( 0 )  1, y ( 0 )  2
Taking the Laplace transform on both sides:
2
s Y (s )  sy ( 0 )  y ( 0 )  3sY (s )  3y ( 0 )  2Y (s )  5 /s
Substituting the values of the initial conditions
into the last equation. Then solving for Y(s) and
expanding by partial-fraction, we get
s 2  s  5 5 5 3
Y (s )    
s ( s  1) ( s  2 ) 2s s 1 2 (s  2)

Taking the inverse Laplace transform, we obtain

t 2 t
y (t )  2 .5  5e  1 .5e t 0
Example. Solve the following differential equation:

y
(3)
 3y   3y   y  1, y (0)  y ( 0 )  y ( 0 )  0

Solution:
1 b1 b2 b3 c4
F (s )     
s (s  1) 3
s 1 (s  1) 2
(s  1) 3
s

1
b3  [( s  1) 3
]s 1  1
s (s  1) 3

d 1  d 1
b2  [ (s  1) 3
] [ ( ) ]s 1
 d s s ( s  1)
3
s 1 d s s

 ( s 2 )  1
s 1
 1 3
b1  ( 2s )  1
2! s 1

1
c  s 1
s (s  1) 3
s 0

1 1 1 1
F (s )    
( s  1) ( s  1) s 1
3 2
s

1 t
y 1  t e
2
 te t  e t , t 0
2