You are on page 1of 49

Technische Hochschule

Deggendorf

Lecture: Advanced Modeling and Simulation


Chapter: Mathematical Models of Physical Systems

Last Update: 15/03/2017 17:18:00

Authors: Prof. Dr.-Ing. Peter Firsching;Prof. Dr. László Juhász


Prof. Dr. László Juhász

Block Diagram of an electrical excited DC Machine


(taken from Schlitt H.: Regelungstechnik)
Advanced Modeling and Simulation - Mathematical Models of Physical Systems
Last Update: 15/03/2017 17:18:00

Content:

1  DIFFERENTIAL EQUATIONS OF PHYSICAL SYSTEMS 4 


1.1  Common Aspects of Physical Models 4 
1.1.1  Gradient Law 4 
1.1.2  Storage Law 5 
1.1.3  Law of Continuity / Balance Law 5 
1.1.4  Domain-specific laws and methods 5 

1.2  Examples 7 
1.2.1  Electrical System 7 
1.2.2  Mechanical System 7 
1.2.3  Hydraulic System 7 

2  LINEAR APPROXIMATION OF NON LINEAR SYSTEM EQUATIONS 8 


2.1  Taylor Series Expansion 8 
2.1.1  Linearisation of Functions depending on one variable 8 
2.1.2  Linearisation of Functions depending on two or more variables 9 
2.1.3  Example – Characteristic Function of a Valve 9 

2.2  Linearisation of Differential Equations 10 

3  BLOCK DIAGRAM MODELS 11 


3.1  Graphic symbols 11 

3.2  Example – Block Diagram of a Permanent Excited DC Machine 13 

4  STATE VARIABLE MODELS OF LINEAR SYSTEMS 14 


4.1  Different Ways of System description 14 

4.2  State variables and state equations 15 


4.2.1  Definitions 15 
4.2.2  State and output equations for a linear system 15 
4.2.3  Canonical Forms of State Variable Models 17 
4.2.4  Linear transformation of the state equations 20 

4.3  Solution of the state equation 21 


4.3.1  Solution in time domain 21 
4.3.2  Solution in frequency domain 22 

5  DISCRETE TIME SYSTEMS 23 


5.1  The Basic Problem 23 

5.2  Brief Explanation of the Z Transform 24 

© Prof. Dr.-Ing. Peter Firsching;Prof. Dr. László Juhász page 2


Advanced Modeling and Simulation - Mathematical Models of Physical Systems
Last Update: 15/03/2017 17:18:00
5.3  Difference Equations and Z Transfer Functions 25 
5.3.1  Description of Discrete Time Systems 25 
5.3.2  Numerical Integration 26 

6  EXERCISES 29 
6.1  Questions 29 

6.2  Some Solutions 39 

ADDENDUM: LAPLACE TRANSFORM – A SHORT INTRODUCTION 46 

© Prof. Dr.-Ing. Peter Firsching;Prof. Dr. László Juhász page 3


Advanced Modeling and Simulation - Mathematical Models of Physical Systems
Last Update: 15/03/2017 17:18:00

1 Differential Equations of Physical Systems


1.1 Common Aspects of Physical Models
We can find dynamic behaviour in different types of physical systems:
 electrical
 mechanical (translation, rotation)
 thermal
 fluidical (e. g. oil, water, air).

The description of the dynamic behaviour of a system is based on three general


laws. Furthermore, domain-specific methods can be also used (like Thevénin-
theorem for electrical circuits, Lagrange-equations for mechanical systems and so
on).

1.1.1 Gradient Law

In general:
1
q1 (t )  [u (t )  x(t )]
R
R resistance
q1(t) flow value
u(t) – x(t) „gradient“

Examples:

 Electric current (law of Ohm): 1


i (t )  [u1 (t )  u 2 (t )]
R
 mechanical movement 1
(translational, rotational) a (t )  [ F1 (t )  F2 (t )]
m
1
 (t )  [ M 1 (t )  M 2 (t )]
J
 Thermal flow 1
i w (t )  [1 (t )  2 (t )]
R
 hydraulic flow (law of Hagen- 1
Poiseuille) q (t )  [ p1 (t )  p2 (t )]
RP

© Prof. Dr.-Ing. Peter Firsching;Prof. Dr. László Juhász page 4


Advanced Modeling and Simulation - Mathematical Models of Physical Systems
Last Update: 15/03/2017 17:18:00

1.1.2 Storage Law

A flow value x(t) will be accumulated in a storage medium.


differential way: q 2 ( t )  K x (t )
t

integral way: x(t )  x(0)   q 2 ( )d


0

1.1.3 Law of Continuity / Balance Law

Only two values: q2 (t )  q1 (t )


n

general node equation:  q (t )  0


i 1
i

This law is the result of balance equations for the specific type of system:

 Electrical -> mesh or node equations (Kirchhoff)


 Mechanical -> balance of forces / torques
 Thermal -> balance of thermal flows
 Fluidical -> balance of pressure / volume flow

1.1.4 Domain-specific laws and methods

Mechanical systems:

- Potential and kinetic energies of masses

- Potential energy stored in an elastic spring

- Lagrange equations of mechanical systems

© Prof. Dr.-Ing. Peter Firsching;Prof. Dr. László Juhász page 5


Advanced Modeling and Simulation - Mathematical Models of Physical Systems
Last Update: 15/03/2017 17:18:00
- Use of relative movements and coordinate differences when the complete
difference path is zero.

© Prof. Dr.-Ing. Peter Firsching;Prof. Dr. László Juhász page 6


Advanced Modeling and Simulation - Mathematical Models of Physical Systems
Last Update: 15/03/2017 17:18:00

1.2 Examples
1.2.1 Electrical System

1 Simple RC network
Potential law: i1 (t )  (u1 (t )  u 2 (t ))
R
i1(t)
 1
Storage Law: u 2 (t )  i2 (t ) i2(t)
C R
u1(t) u2(t)
Balance Law: i1 (t )  i2 (t )  0 C

Result:

Figure 1
RC u 2 (t )  u 2 (t )  u1 (t )

1.2.2 Mechanical System

Potential law: Fc (t )  c ( x  x0 ) Combination of linear Spring



and Damper
Storage Law: Fd (t )  d x (t )
x0 d x
Balance Law: F (t )  Fc (t )  Fd (t )  0 Fd(t)
F(t)
Result:
Fc(t)
d  1
x(t )  x(t )  F (t ) c
c c Figure 2

1.2.3 Hydraulic System

Potential law: Q2 (t )  A 2 gh Tank with inflow and outflow



Storage Law: QT (t )  A h(t )
Balance Law: Q1 (t )  Q2 (t )  QT (t )  0 Q1(t)

Result: h QT(t)
1  1
h(t )  h(t )  Q1 (t ) Q2(t)
2g A 2g
Figure 3

© Prof. Dr.-Ing. Peter Firsching;Prof. Dr. László Juhász page 7


Advanced Modeling and Simulation - Mathematical Models of Physical Systems
Last Update: 15/03/2017 17:18:00

2 Linear Approximation of Non Linear System


Equations
2.1 Taylor Series Expansion

The linearisation of a non linear function is the result of a Taylor Series


Expansion where the higher order terms are neglected.

2.1.1 Linearisation of Functions depending on one variable

Given a non linear function Y = F(X).

 Taylor series expansion at a working point P(X0, Y0):


2
d 1 d  2
Y  Y0  f (X ) (X  X 0 )    f ( X ) ( X  X 0 )    Rn 1 ( X m )
2

dX X0 2!  dX 
X 0

 For a small excursion x = X – X0 about P(X0, Y0) we can find an approximation


d d
Y  Y0  f ( X ) ( X  X 0 )  Y0  f ( X ) x
dX X0 dX X0

 Linearisation of the function Y = f(X):

d
y  Y  Y0  f ( X ) x
dX X0

© Prof. Dr.-Ing. Peter Firsching;Prof. Dr. László Juhász page 8


Advanced Modeling and Simulation - Mathematical Models of Physical Systems
Last Update: 15/03/2017 17:18:00

2.1.2 Linearisation of Functions depending on two or more variables

Given a non linear function Y = F(X1, X2, ...., Xn) = F(X)

 Taylor series expansion at a working point P(X10, X20, ..., Xn0,Y0):


n
f ( X 0 ) 1 n  2
Y  Y0   [( X i  X i 0 ) ]   [( X i  X i 0 ) ] f ( X 0 )    Rn1 ( X m )
i 1 X i 2! i 1 X i

 For a small excursion x = X – X0 about P(X0, Y0) we can find an approximation:


n
f ( X 0 ) n
f ( X 0 )
Y  Y0   [( X i  X i 0 ) ]  Y0   [ xi ]
i 1 X i i 1  X i

 Linearisation of the function Y = f(X):


n
f ( X 0 )
y  Y  Y0  [ xi ]
i 1 X i

2.1.3 Example – Characteristic Function of a Valve

Figure 4 shows a simple valve which is controlled by a slider mechanism. The


calculation of the output volume flow is nonlinear function depending on the slider
position YN and the load pressure pL.

pL YN
QV  Q0YN 1 
p0 p0 pL QV(t)

QV: volume flow through the valve


Q0: maximum volume flow Figure 4
YN: standardised slider position
p0: supply pressure
pL: load pressure

The linearisation of the characteristic function of the valve at a small excursion about
a working point P0(YN0, pL0) yields a linear expression:

QV QV
QV  QV  QV 0  YN  p L  K y YN  K p p L
YN p L Y
YN 0 , pL 0 N 0 , pL 0

© Prof. Dr.-Ing. Peter Firsching;Prof. Dr. László Juhász page 9


Advanced Modeling and Simulation - Mathematical Models of Physical Systems
Last Update: 15/03/2017 17:18:00

2.2 Linearisation of Differential Equations


When the description of a dynamic system contains nonlinear expressions we must
solve the problem of the linearization of a differential equation. The solution of this
problem should be explained by the following example.

Given a nonlinear homogenous differential equation


 
x (t )  2 x(t )  cos x  0
The differential equation should be linearised for a small excursion about x0 = /4.

1. Since x = x - x0 we can solve for x = x0 + x = /4 + x


d2  d
(  x )  2(  x)  cos(  x)  0
dt 2 4 dt 4 4

2. Calculation of the time derivatives yields


d2  
(   x )     x
dt 2 4
d  
(  x)     x
dt 4
3. The expression cos(/4 + x) can be linearized by a truncated Taylor series
d
cos(  x)  cos( )  cos x ( x   )  cos( )  sin( ) x
4 4 dx 4 4 4
x0 
4
1 1
 cos(  x)  2 2x
4 2 2
4. Applying all calculated term to the differential equation yields

  1 1
x(t )  2 x(t )  2x   2
2 2

© Prof. Dr.-Ing. Peter Firsching;Prof. Dr. László Juhász page 10


Advanced Modeling and Simulation - Mathematical Models of Physical Systems
Last Update: 15/03/2017 17:18:00

3 Block Diagram Models


3.1 Graphic symbols
Block diagrams represent a mathematical formula using graphic symbols. For the
visualisation of differential equations the following simple symbols are sufficient.

List of simple symbols:

u3 Summation
Negative sign is placed at a summation input where necessary.
u1
y y (t )  u1 (t )  u 2 (t )  u3 (t )
u2

Constant

u
K
y
y (t )  K u (t )

Ki
Integration
Ki is the integration constant
t
u y
y (t )  K i  u ( ) d  y0
0

Boxes with a double border often represent non linear


y=f(u) functions
u y
 hysteresis
 slip stick effects,
 characteristic functions.

u3 Multiplication in general
y (t )  u1 (t ) u 2 (t ) u3 (t )
u1 y
u2

Boxes with integrated functions

K, T
Functions which are often used in complex systems can
be designed as specific boxes.
u y 
Example: PT1 system y (t )  T y (t )  K u (t )

© Prof. Dr.-Ing. Peter Firsching;Prof. Dr. László Juhász page 11


Advanced Modeling and Simulation - Mathematical Models of Physical Systems
Last Update: 15/03/2017 17:18:00

Block Diagram Transformations


Block diagram transformations are arithmetic rules for the graphic representation of
formulas.

Figure 5: taken from [DOB]

© Prof. Dr.-Ing. Peter Firsching;Prof. Dr. László Juhász page 12


Advanced Modeling and Simulation - Mathematical Models of Physical Systems
Last Update: 15/03/2017 17:18:00

3.2 Example – Block Diagram of a Permanent Excited DC Machine


A permanent excited DC machine can be described by the following set of equations:

 armature circuit
LA  1
i A (t )  i A (t )  [u A (t )  e A (t )]
RA RA
 Inner torque
M M (t )  K 2 i A (t )
 mechanic behaviour
 1
n(t )  [ M M (t )  M L (t )]
2J
t
1
n(t )   [ M M ( )  M L ( )]d  n0
0
2 J
 Back electromotive force
e A (t )  K1 n(t )

The equations can be directly interpreted as a block diagram structure shown in


figure 6.

ML(t)
1/RA LA/RA 1/2J

K2
uA(t) n(t)
iA(t) MM(t)
eA(t)

K1

Figure 6: Block Diagram DC Machine

© Prof. Dr.-Ing. Peter Firsching;Prof. Dr. László Juhász page 13


Advanced Modeling and Simulation - Mathematical Models of Physical Systems
Last Update: 15/03/2017 17:18:00

4 State Variable Models of Linear Systems


4.1 Different Ways of System description
The description of the behaviour of a system output y against a system input u is
usually given by

 an nth order differential equation  time domain


 a transfer function (Laplace function) in case of a linear  frequency domain
system

Example: output values of a RLC network

Given a simple RLC network in Figure 7.

iL(t)
R
u1(t) L C uC(t)

Figure 7: RLC Network

We can calculate the behaviour of the voltage uc against u1 and the behaviour of the
current iL against u1 in time and frequency domain. Thus we get two different
system models.

Time Domain Frequency Domain

Voltage uC(t) Voltage UC(s)


  
U C (s) sL
LRC uc (t )  L uc (t )  R uc (t )  L u1 (t ) Fu ( s )   2
U 1 ( s ) s LRC  sL  R

Current iL(t) Current IL(s)


  I L (s) 1
LRC iL (t )  L iL (t )  R iL (t )  u1 (t ) FI ( s)   2
U 1 ( s ) s LRC  sL  R

Can we find a system model which describes the behaviour of more than one system
output in a compact form? The problem can be solved by a state variable model of
this network in the following chapter.

© Prof. Dr.-Ing. Peter Firsching;Prof. Dr. László Juhász page 14


Advanced Modeling and Simulation - Mathematical Models of Physical Systems
Last Update: 15/03/2017 17:18:00

4.2 State variables and state equations


In this chapter we briefly discuss the basics of state variable models of dynamic
systems.

4.2.1 Definitions

 What are state variables? (from [NSN], pp. 134 ...)


State variables are the smallest set of linearly independent system variables such
that the values of the members of the set at a time t0 along with known forcing
functions completely determine the value of all system variables for all t >= t0.

 What is a state vector? (from [NSN], pp. 134 ...)


A state vector is a vector whose elements are the state variables.

 What is the state space? (from [NSN], pp. 134 ...)


The state space is an n dimensional space whose axis are the state variables.

 What are the state equations? (from [NSN], pp. 134 ...)
The state equations are a set of n simultaneous first-order differential equations
with n variable, where the n variables to be slved are the state variables.

 What are the output equations? (from [NSN], pp. 134 ...)
The output equations are a set of algebraic equations that expresses the output
variables of a system as a linear combination of the state variables and the
inputs.

4.2.2 State and output equations for a linear system

A linear system with k inputs, m outputs and n state variables can be described by a
set of state equations and an output equation in vector matrix form.


x  A x B u
n x1 n x n n x1 n x k k x1

y  C x D u
m x1 m x n n x1 m x k k x1

x state vector A system matrix


x derivative of the state vector B input matrix
with respect to time
y output vector C output matrix
u input vector / control vector D feedforward matrix

© Prof. Dr.-Ing. Peter Firsching;Prof. Dr. László Juhász page 15


Advanced Modeling and Simulation - Mathematical Models of Physical Systems
Last Update: 15/03/2017 17:18:00

In the special case of a system with a single input and a single output the state
and output equations can be written as follows. This system is also called a SISO
system.


x  A xbu
T
y c xd u

How to determine the minimum number of state variables?

The number of the state variables is determined by the order of the differential
equation describing the dynamic system.

Example: state and output equations of a RLC network


In chapter 4.1 we found the 2nd order system model of a RLC network. With
uc(t) and iL(t) as state variables we get a state model of the RLC network. The
system outputs are the state variables

State Equation:
 1 1
      1 
u C    RC C  uC   
1   RC  u1
 i L   0   L   0 
i 
 L 
Output Equation:
uC  1 0 uC 
 i   0 1   i 
 L   L
The structure of the physical system contains two energy storing element. The
capacity can store electric energy (W = ½ C U2). The inductance can store magnetic
energy (W = ½ L I2). In many cases we can determine the number and the choice of
state variable directly from the energy storing elements in the physical system. To get
the right number of state variables we must make sure that the energy storing
elements are linear independent. In other words we can say that it does not exist
any linear combination to calculate a state variable depending from all other state
variables of a system with coefficient unequal zero.

© Prof. Dr.-Ing. Peter Firsching;Prof. Dr. László Juhász page 16


Advanced Modeling and Simulation - Mathematical Models of Physical Systems
Last Update: 15/03/2017 17:18:00

Examples: physical systems and the system order

One energy storing element (capacity C)


-> first order system
R
u1(t) C uC(t)

Two linear independent energy storing


elements (capacities C1 and C2)
R1 R2
u1(t) C1 C2 uC(t) -> second order system

Two linear independent energy storing


iL(t) elements (capacity C and inductance L)
R
u1(t) L C uC(t) -> second order system

Three energy storing elements (C1, C2


and C3). The voltage at one capacity
R1
u1(t) C1
C2
C3 uC(t) can be expressed by a linear
combination of the other capacity
voltages (mesh equation).
Thus the system order is not three but
two.

4.2.3 Canonical Forms of State Variable Models

In chapter 4.2.2 we saw that there can be close dependencies between the physical
structure of a system and the state variables. But it is also possible to calculate a
state variable model of a system only based on the behaviour of the system output
against the system input.

If given the transfer function of a system we can easily calculate a state variable
model with a simple internal structure. Since in most technical systems the feed
forward path does not exist we can define the transfer function as

N ( s ) bm s m  bm1s m1    b1s  b0


F (s)   n ( n  m)
D( s) s  an1s n1    a1s  a0

© Prof. Dr.-Ing. Peter Firsching;Prof. Dr. László Juhász page 17


Advanced Modeling and Simulation - Mathematical Models of Physical Systems
Last Update: 15/03/2017 17:18:00

The transfer function can be separated into two transfer functions

X ( s) Y ( s) 1
F ( s )  F1 ( s ) F2 ( s )   N (s)
U ( s) X ( s) D( s)

The transformation of F1 back to time domain yields:

X ( s) 1
F1 ( s )    U ( s)  D( s) X ( s)
U ( s) D( s)
 L1{U ( s )}  L1{D( s ) X ( s )}

 u (t )  x ( n ) (t )  an1 x ( n1) (t )    a1 x(t )  a0 x(t )

 d x
( )
Using the substitution x  x 1  x   ( = 1 .. n-1) we can find a state
dt
representation of the differential equation:

      0 1 0  0 
 x  0 
 x 1   x   0 0 1  0   1   
x2   x   x 0
          2    u (t )
       0      
 0 0 1    
 x n   x ( n)   x 1
     a0  a1  a2   an1   n   

The output equation can be calculated from F2(s).


Y ( s)
F2 ( s )   N ( s)  Y ( s)  N ( s) X ( s)
X ( s)
 L1{Y ( s )}  L1{N ( s ) X ( s )}

 y (t )  bm x ( m ) (t )  bm1 x ( m1) (t )    b1 x(t )  b0 x(t )

 x1 
x 
y (t )  b0 b1 b2  bm  0  0  2

 
 xn 

© Prof. Dr.-Ing. Peter Firsching;Prof. Dr. László Juhász page 18


Advanced Modeling and Simulation - Mathematical Models of Physical Systems
Last Update: 15/03/2017 17:18:00

The structure of the system is shown in figure 8. This structure is also called
controllable canonical form.

bm

bm-1

b0
u(t) y(t)
xn xn-1 x1
-an-1

-an-2

-a0
Figure 8: Structure of the Controllable Canonical Form

There are two other canonical structures of state space models which are:
 the observable canonical form
 the diagonal form

The characterstics of the two additional canonical structures are shown in the
following figures 9 and 10.

Observable canonical form:


bm

b1

x1
b0
u(t) xn xn-1 y(t)
-an-1

-a1

-a0

Figure 9: Observable Canonical Form

© Prof. Dr.-Ing. Peter Firsching;Prof. Dr. László Juhász page 19


Advanced Modeling and Simulation - Mathematical Models of Physical Systems
Last Update: 15/03/2017 17:18:00

x1
b1* c1

s1

xn-1
u(t) bn-1* cn-1 y(t)

sn-1

xn
bn* cn

sn

Figure 10: Diagonal Form

4.2.4 Linear transformation of the state equations

Applying a regular transformation


x T z

any kind of state space description of a linear system can be computed.

The new state equations are


* *
z  A z  B u
* *
yC zD u

with
* 1 * 1 * *
A T A T; B  T B; C  C T ; D  D .

This kind of transformation can be used to achieve the diagonal canonical form of the
matrix A (Jordan canonical form).

© Prof. Dr.-Ing. Peter Firsching;Prof. Dr. László Juhász page 20


Advanced Modeling and Simulation - Mathematical Models of Physical Systems
Last Update: 15/03/2017 17:18:00

4.3 Solution of the state equation

4.3.1 Solution in time domain


In time domain the solution of the state equation must be divided into two steps. In
the first step we must solve the homogeneous part of the state equation

xAx (u  0, u  0)

Since the state equation is a set of first order differential equations the solution yield
x(t )   (t  t 0 ) x(t 0 )  e A ( t t0 ) x (t 0 ) (t  t 0 )

(t) is called the transition matrix and can be defined by an infinite series
At t2 3 t
3
2
 (t )  e  I  At  A A 
2! 3!
In the second step the solution of the inhomogeneous state equation can be
generated by well known mathematical methods of solving differential equation.

We define a general solution of the state equation


x(t )  e A t f (t )

Applying this solution to the inhomogeneous state equation yield


  
x(t )  A e At
f (t )  e At
f (t ) with e At
f (t )  B u (t )

The term f(t) can be calculated by integrating the differential equation



f (t )  e  A t B u (t )

Applying the solution of f(t) to the general solution of the state equation yield

e
 A
x(t )  e At
B u ( )d
t0

This is the particulate solution of the inhomogeneous state equation with u(t0) = 0.

The complete solution is a superposition of the homogeneous and the particulate


part. Using the abbreviation (t) we get the final solution

t
x(t )  (t  t 0 ) x(t 0 )   (t   ) B u ( )d
t0

© Prof. Dr.-Ing. Peter Firsching;Prof. Dr. László Juhász page 21


Advanced Modeling and Simulation - Mathematical Models of Physical Systems
Last Update: 15/03/2017 17:18:00

4.3.2 Solution in frequency domain

The solution of the state equation in the frequency domain is based on the Laplace
transform.

Given a state variable model of a linear SISO system:



x  A xbu
T
yc x

Transforming both equation to the Laplace space and solving for y yields:
T
Y ( s )  c ( s I  A) 1 b U ( s )

Now we can calculate a transfer function F(s) = Y(s) / U(s)

Y ( s) 1 T
F (s)   c ( s I  A) adj b
U ( s ) det( s I  A)

The transfer function can now be used to calculate the signal response functions in
the frequency domain.

In case of a system with multiple inputs and multiple we get k * m different transfer
functions (k = number of inputs, m = number of outputs). The transfer function can be
combined in a transfer matrix F(s).

 F11 F12  F1k 


F F22  F2 k 
Y ( s )  F ( s) U ( s )   21 U (s)
     
 
 Fm1 Fm 2  Fmk 

© Prof. Dr.-Ing. Peter Firsching;Prof. Dr. László Juhász page 22


Advanced Modeling and Simulation - Mathematical Models of Physical Systems
Last Update: 15/03/2017 17:18:00

5 Discrete Time Systems


5.1 The Basic Problem
Discrete time systems occur when input and / or output signals of continuous time
system are for example processed with a digital system (computer, DSP etc.). In
figure 11 is shown how a continuous time system is sampled with a constant
sampling time period T.

f(t) f(kT)
fk

sampling

T 2T 3T t T 2T 3T kT

Figure 11: example of signal sampling

In case of a constant sampling time T the sampled signal can be interpreted as a


series of unit impulses 0(t).


f ( kT )  f ( k )  f * (t )   f k  0 (t  kT )
k 0

The unit impulse is defined as:


  0 (t )  0 for all t0

 

0 (t ) dt  1

© Prof. Dr.-Ing. Peter Firsching;Prof. Dr. László Juhász page 23


Advanced Modeling and Simulation - Mathematical Models of Physical Systems
Last Update: 15/03/2017 17:18:00

5.2 Brief Explanation of the Z Transform

Definition of the Z Transform

Since we can describe a discrete time signal or a sampled discrete time function like
a continuous time signal / function it is possible to calculate the Laplace transform of
this signal / function.

f (t ) o    
*
F ( s )   f k e  s kT
*

k 0

We substitute
z  e sT
and thus we get

*
F (s)  F ( z )   f k z k
z  e sT
k 0

F * ( s) is the called the Z transform of the impulse series f*(t) = f(kT)


z e sT

Possible notations for the Z Transform:


 F(z) = Z{ f(kT) }
 f(kT) o---- F(z)

Mapping of the s Plane to the z Plane

The mapping from s plane to z plane is given by the expression


z  e sT  e (  j )T  eT e jT
 In case of   0 the function z = esT maps the left s half plane into the interior of a
unit circle
 In case of  = 0 the function z = esT maps the imaginary axis of the s half plane to
the unit circle itself.
s plane j Im{s} z plane j Im{z}

Mapping

Re{s} Re{z}

Figure 12: mapping from s plane to z plane

© Prof. Dr.-Ing. Peter Firsching;Prof. Dr. László Juhász page 24


Advanced Modeling and Simulation - Mathematical Models of Physical Systems
Last Update: 15/03/2017 17:18:00

Properties of the Z Transform


For further considerations we should know two special properties of the Z transform.

 Right shift in time


1
Z { f ((k  k 0 )T )}  F ( z)
z k0
 Left shift in time
Z { f (( k  k 0 )T )}  z k0 F ( z )

5.3 Difference Equations and Z Transfer Functions


5.3.1 Description of Discrete Time Systems
Let be given a linear time invariant discrete time system. The behaviour of this kind of
system can be described by a difference equation:

y ( k )  a n 1 y (k  1)    a1 y (k  n  1)  a 0 y ( k  n) 
bn u (k )  bn 1u (k  1)    b1u (k  n  1)  b0 u (k  n)

Applying the right shift rule (see chapter 5.2) hence we get a z transform
corresponding to the difference equation

Y ( z ) bn z n  bn 1 z n 1    b1 z  b0
F ( z)  
U ( z) z n  a n 1 z n 1   a1 z  a 0

The expression z-1 in the discrete frequency domain can be interpreted as a dead
time element with dead time T (sampling time) in the time domain.

If we want to sketch the block diagram structure of the transfer function F(z) we need
a new block diagram symbol which is shown in figure 13.

u(k)
z-1
y(k)
y (k )  u (k  1)

Figure 13: block diagram of a dead time element

© Prof. Dr.-Ing. Peter Firsching;Prof. Dr. László Juhász page 25


Advanced Modeling and Simulation - Mathematical Models of Physical Systems
Last Update: 15/03/2017 17:18:00

The block diagram structure of F(z) is shown in figure 14.

u(k) u(k-1) u(k-2) u(k-n)


z-1 z-1 z-1 z-1

bn bn-1 bn-2 bn-3 b0

y(k)

a0 a1 a2 an-1

z-1 z-1 z-1 z-1


y(k-n) y(k-n+1) y(k-1)

Figure 14: block diagram structure of a general discrete time system

5.3.2 Numerical Integration

The basic principle of the numerical integration of a function y = f(x) is the stepwise
approximation of the area below the function. A well known algorithm of numerical
approximation is the method of Euler (German Mathematician, 1707 – 1783). Figure
15 illustrates how the approximation of the area below a function can be done using
rectangle areas.
f(x) F(x)

F(0)
INTEGRATION

x x
1 2 3 4 0 1 2 3 4

x
F ( x)   f (  ) d  F ( 0)
x 0

Figure 15: Illustration of the Method of Euler

For the next step we assume that


 the function f(x) can be interpreted as function of time f(t)
 the time steps t  t k  t k 1 are all equal. Thus we can say: t = T.

© Prof. Dr.-Ing. Peter Firsching;Prof. Dr. László Juhász page 26


Advanced Modeling and Simulation - Mathematical Models of Physical Systems
Last Update: 15/03/2017 17:18:00

Using the abbreviations F(kT) = y(k) and f(kT) = x(k) the method of Euler can now be
described by a difference equation.

y ( k )  y ( k  1)  x( k  1) * T

This difference equation can be transformed to the z domain.

Y ( z )  z 1Y ( z )  z 1TX ( z )
Y ( z )( z  1)  TX ( z )

Transfer function of the numerical Integration using the method of Euler

Y ( z) 1
FEuler ( z )  T
X ( z) z 1

T
“1/s” in the Laplace domain can be approximately expressed by in z domain
z 1
Thus we can use the transfer function FEuler to convert a Laplace transfer function
F(s) to the z domain.

Example

Given a proportional system with a first order time delay (PT1)


1
F ( s) 
1  sT1

1 T
Applying the approximation  we get a z transfer function
s z 1

T
F ( z) 
(T  T1 )  T1 z

Figure 16 shows a comparison between the step response function of the two
transfer functions F(s) and F(z) with T = 0.1 s and T1 = 1 s.

© Prof. Dr.-Ing. Peter Firsching;Prof. Dr. László Juhász page 27


Advanced Modeling and Simulation - Mathematical Models of Physical Systems
Last Update: 15/03/2017 17:18:00

Figure 16: continuous time and discrete time step response of a PT1 system

© Prof. Dr.-Ing. Peter Firsching;Prof. Dr. László Juhász page 28


Advanced Modeling and Simulation - Mathematical Models of Physical Systems
Last Update: 15/03/2017 17:18:00

6 Exercises
6.1 Questions
1. Sketch the block diagrams of the following systems given by three differential or
integral equations.

a) T y (t )  y (t )  0
 
b) y (t )  2 D 0 y (t )   0 y (t )  ku (t )
2

t
1
2J 0
c) n(t )  M ( )d

2. Figure 17 shows a simple speed controller using a propeller. The driving force at
the rope role (in figure 17: Seilrolle) with radius r is Fa. The air resistance torque is
given by
M R  kP  2 (k P  0,   2n)
The inertia of all rotating elements is J.

Figure 17

a) Find the differential equation for the system. The angular speed (t) is the
system output. Fa(t) is the system input.

b) Sketch the block diagram of the system.

c) Calculate the linearization of the differential equation at the working point


P0(FA0, 0). Derive the linear differential equation for small excursions  and
Fa .

d) Sketch the block diagram of the linearized system.

© Prof. Dr.-Ing. Peter Firsching;Prof. Dr. László Juhász page 29


Advanced Modeling and Simulation - Mathematical Models of Physical Systems
Last Update: 15/03/2017 17:18:00

3. Given a mechanical system in figure


18. The system consists of a linear x rope roll
d
damper (damping constant d) and a
discrete mass m. The rope roll is
supposed to be free of inertia.

Figure 18
m

a) Derive the dynamic behaviour of the mechanic system relating to the


coordinate x by a differential equation. The force Fg = m g is the system input.

b) Derive the behaviour of the velocity x (t )  v(t ) of the damper in case of an


abruptly (like a step) action of the mass on the damper. Graph the behaviour
of v(t) in a diagram and characterise the typical parameters of the function
graph. Specify the type of the system.
V ( s)
c) Determine the transfer function F ( s)  of the system.
Fg ( s)

4. Given a mechanical system in figure


19. The system consists of a linear d x
damper (damping constant d), a linear
spring (spring constant c) and a F
discrete mass m. c m

Figure 19 no friction

a) Derive the dynamic behaviour of the mechanic system related to the


coordinate x by a differential equation. The force F(t) is the system input.

b) Find a state variable model of this system using x(t) and x (t )  v(t ) as state
variables.

5. Given a mechanical system in figure


20. The system consists of two linear
springs (spring constant k1,2) and two
discrete masses m1,2.

Figure 20

Derive the dynamic behaviour of the mechanic system related to the coordinates
x1,2 by a differential equation. The force f(t) is the system input.

© Prof. Dr.-Ing. Peter Firsching;Prof. Dr. László Juhász page 30


Advanced Modeling and Simulation - Mathematical Models of Physical Systems
Last Update: 15/03/2017 17:18:00
6. Given a mechanical system in figure
21. The system consists of a linear
dampers (damping constant c2) two
linear springs (spring constants k1,2)
and two discrete masses m1,2.

Figure 21

Derive the dynamic behaviour of the mechanic system related to the coordinates
x1,2 by a differential equation. The force f(t) is the system input.

7. Given a mechanical system in figure


22. The system consists of two linear
dampers (damping constants c1,2)
linear springs (spring constants k1,2)
and two discrete masses m1,2.

Figure 22

Derive the dynamic behaviour of the mechanic system related to the coordinates
x1,2 by a differential equation. The force f(t) is the system input.

8. Given a mechanical system in figure


23. The system consists of two linear
dampers (damping constants c1,3)
linear springs (spring constants k1,2)
and two discrete masses m1,2.

Figure 23

Derive the dynamic behaviour of the mechanic system related to the coordinates
x, z by a differential equation. The force f(t) is the system input.

© Prof. Dr.-Ing. Peter Firsching;Prof. Dr. László Juhász page 31


Advanced Modeling and Simulation - Mathematical Models of Physical Systems
Last Update: 15/03/2017 17:18:00
9. Given a mechanical system in figure
24.

Derive the dynamic behaviour of the


mechanic system related to the
coordinates x1,2,3 by a differential
equation. The force f(t) is the system
input.
Figure 24

10. Given a mechanical system in figure


25.

Derive the dynamic behaviour of the


mechanic system related to the
coordinates x1,2 by a differential
equation.
Figure 25

11. Given a mechanical system in figure


26.

Derive the dynamic behaviour of the


mechanic system related to the
coordinates x1,2 by a differential
equation. The force f(t) is the system
input.
Figure 26

© Prof. Dr.-Ing. Peter Firsching;Prof. Dr. László Juhász page 32


Advanced Modeling and Simulation - Mathematical Models of Physical Systems
Last Update: 15/03/2017 17:18:00
12. Given a mechanical system in figure
27.

Derive the dynamic behaviour of the


mechanic system related to the
coordinates x1,2 by a differential
equation. The force f(t) is the system
input.
Figure 27

13. Given a mechanical system in figure


28.

Derive the dynamic behaviour of the


mechanic system related to the
coordinate θ by a differential equation.
The force f(t) is the system input.
Figure 28

© Prof. Dr.-Ing. Peter Firsching;Prof. Dr. László Juhász page 33


Advanced Modeling and Simulation - Mathematical Models of Physical Systems
Last Update: 15/03/2017 17:18:00
14. Given a mechanical system in figure
29.

Derive the dynamic behaviour of the


mechanic system related to the
coordinates x1,2 by a differential
equation. The force f(t) is the system
input.
Figure 29

15. Given a mechanical system in


figure 30.

Derive the dynamic behaviour


of the mechanic system
related to the coordinates x1
and θ by a differential
equation. The force f(t) is the
system input.
Figure 30

16. Given a mechanical system in figure


31.

Derive the dynamic behaviour of the


mechanic system related to the
coordinates θ and x1,2 by a differential
equation. The force f(t) is the system
input.
Figure 31

17. Simple Pendulum

© Prof. Dr.-Ing. Peter Firsching;Prof. Dr. László Juhász page 34


Advanced Modeling and Simulation - Mathematical Models of Physical Systems
Last Update: 15/03/2017 17:18:00

a) Find the differential equation describing the


angular movement (t) of the pendulum.

b) Calculate the linearisation of the system at


P0(0 =0°).

c) Find a state space model of this system. 

18. Given a thermal system in figure 32.


Obtain the dependency of the
temperature in the chambers on the
input heat qi(t).

Figure 32

19. Given a thermal system in figure 33.


Obtain the thermal resistance of the
sandwich wall consisting of two layers
with individual thermal resistances of
R1 and R2.

Figure 33

20. Given a thermal system in figure 34


consisting of a water reservoir and an
electrical heater. The water
temperature in the reservoir is
homogenous and the water flow is
constant. Obtain the model of the
temperature change (θ) in the
reservoir.
Figure 34

21. Filling level of a reservoir


The varying level of a fluid in a tank should be calculated (figure 34). The input
flow Qin is constant. The outlet is at the bottom of the reservoir. The speed of the
output flow is v(t )  2 gh(t ) .

© Prof. Dr.-Ing. Peter Firsching;Prof. Dr. László Juhász page 35


Advanced Modeling and Simulation - Mathematical Models of Physical Systems
Last Update: 15/03/2017 17:18:00

Width of the reservoir: AT = 1 m2 Qin


Input Flow: Qin = 5 l/s
Outlet Width: AO = 0.01 m2
h(t)

Figure 35
Qout

a) Find the differential equation describing the dynamic behaviour of the filling
level of the reservoir. Use the given values.

b) Sketch the block diagram structure of the system.

c) Calculate the linearisation of the system at P0(h0=0,5 m).

22. Given a hydraulic system


in figure 36. Obtain the
model of the pressure and
level changes in nearby
the nominal levels(use the
linearized model of the
system).
Figure 36

23. Given a hydraulic system


in figure 37. Obtain the
model of the pressure
changes in nearby the
nominal levels (use the
linearized model of the
system).
Figure 37

© Prof. Dr.-Ing. Peter Firsching;Prof. Dr. László Juhász page 36


Advanced Modeling and Simulation - Mathematical Models of Physical Systems
Last Update: 15/03/2017 17:18:00
24. Given the electrical system in figure
38. Obtain the state-space model of
the system.
Figure 38

25. Find the state variable model of a permanent excited DC machine based on the
equations in chapter 3.2.
 Use n(t) and iA(t) as state variables.
 Use uA(t) and ML(t) as input variable.

n( s )
26. Calculate the transfer function F ( s )  of the permanent excited DC machine
u A ( s)
given in chapter 3.2. Use this transfer function to calculate the controllable
canonical form of a state variable model.

27. Given the electro-mechanical system


of a galvanometer shown in figure 39.
Obtain transfer-function of the system.

Figure 39

28. Given the electro-mechanical system


of a microphone shown in figure 40.
Obtain transfer-function of the system.

Figure 40

© Prof. Dr.-Ing. Peter Firsching;Prof. Dr. László Juhász page 37


Advanced Modeling and Simulation - Mathematical Models of Physical Systems
Last Update: 15/03/2017 17:18:00
29. Proof the correctness of the state variable model of the RLC network in chapter
4.2.2

30. Given the following transfer functions. Find the corresponding controllable
canonical form in time domain (state equation and output equation). Sketch the
block diagram structure.
Y (s) s
a) F (s)  
U ( s) s  1
T
Y ( s) K
b) F ( s)   2
U (s) s  2 D  0s   0 2
Y ( s) b1 s  b0
c) F (s)   3
U ( s) s  a 2 s 2  a1 s  a 0

31. Find the state equation and the output equation for the observable canonical form
and the diagonal form given in figures 9 and 10, chapter 4.2.3.

32. Given the following z transfer functions. Find the corresponding difference
equations.
Y ( z ) b1 z  b0
a) F1 ( z )   .
U ( z) z  a0
Y ( z) 5z
b) F2 ( z )   2
U ( z) z  z  6

© Prof. Dr.-Ing. Peter Firsching;Prof. Dr. László Juhász page 38


Advanced Modeling and Simulation - Mathematical Models of Physical Systems
Last Update: 15/03/2017 17:18:00

6.2 Some Solutions


Question 1:

a)

y(t)

1/T

b)
k
u(t) y(t)

2D0

2
0

c)
1/2J
M(t) n(t)

Question 2:

a) J  k p 2  Fa r
b) r kp
  Fa   2
J J

r/J
Fa(t) (t)
2
kp/J x

c) d 2
Linearisation:  2  02   (   0 )   0  2 0 
2

d 0

Linear Differential Equation: J   2 0 k p   Fa r

© Prof. Dr.-Ing. Peter Firsching;Prof. Dr. László Juhász page 39


Advanced Modeling and Simulation - Mathematical Models of Physical Systems
Last Update: 15/03/2017 17:18:00

d)
r/J
Fa(t) (t)

2kp0/J

Question 3:

a) Differential Equation: mx  dx  mg


b) m 1
PT1 system: v  v  Fg ( Fg  mg )
d d
System Parameters: K = 1/d, T = m/d

1 t
Step Response: v(t )  Fg (1  e m / d )
d
v(t)

vmax

t
T
c) 1
F (s)  d
Transfer Function:
1 sm
d
Question 4:

a) Differential Equation: mx  dx  cx  F (t )


b) 1. Equation: x  v
2. Equation: x  v
d c 1
 v  v  x  F (t )
m m m
 x   0 1   x  0 
State Equation:  v    c d      1 F
   m  m  v   m

© Prof. Dr.-Ing. Peter Firsching;Prof. Dr. László Juhász page 40


Advanced Modeling and Simulation - Mathematical Models of Physical Systems
Last Update: 15/03/2017 17:18:00

 x
Output Equation: y  x  1 0 
v 
Question 17:

a)
m2 m 2  m3
Differential Equation: 0.044 h  1m h  0.005
s s
b)
m m
Block Diagram Structure: h  0.005  0.044 h
s s

0.005
Qin(t) h(t)

2
0.044 x

c) d
Linearisation: h  h0  h (h  h0 )
dh h0

1 1
h  h0  H  h
2 h0
1 1
 H  h
2 m

2 0.005
Linear Differential Equation: h  s h  2m
0.044 0.044
Question 21:

a) Differential Equation: J  mg sin   0


J: inertia of the pendulum
b) Linear Differential Equation: J  mg  0

c) State Variable Model: x1  


x2  
mg
   
J

© Prof. Dr.-Ing. Peter Firsching;Prof. Dr. László Juhász page 41


Advanced Modeling and Simulation - Mathematical Models of Physical Systems
Last Update: 15/03/2017 17:18:00

   0 1  
   mg 
0  
   J   

Question 25:

State variable model of a permanently excited DC machine

 K2   1 
0 0 
n 
 2J  n 2J   uA 
i    K1 RA  i   1  M 
 A      A  0   L
 L A L A   L A 

Question 26:

The transfer function can be derived directly from the block diagram structure.
K2
n( s ) 2J LA
F ( s)  
u A ( s) R KK
s2  A s  1 2
LA 2J LA

Thus the controllable canonical form is:

 x1   K K R
0 1 
 x1  0
 x    1 2 A
R 
 A  x   1  u A
 2   2J L A LA   2  

 K L  x 
n(t )   2 A 0  1 
 2J L A   x2 

Question 29:

Calculation of the branch currents:


1. i R  (u1 (t )  u C (t )) / R
1 1
2. iC (t )  u C (t )  (i R (t )  i L (t ))  u C (t )
C C
1
3. iL (t )  u C (t )
L

© Prof. Dr.-Ing. Peter Firsching;Prof. Dr. László Juhász page 42


Advanced Modeling and Simulation - Mathematical Models of Physical Systems
Last Update: 15/03/2017 17:18:00

Thus the state equations are:

1 1 1
 u C (t )   u C (t )  i L (t )  u1 (t )
RC C RC
1
 iL (t )  u C (t )
L

Question 30:

a)  coefficients of the transfer function:


a0 = 1/T, a1 = 1, b1 = 1
 state and output equation:
1
x (t )   x(t )  u (t )
T
1
y (t )   x(t )  u (t )
T
 block diagram structure:

u(t) y(t)
1/T

1/T

b)  coefficients of the transfer function:


a0 = 02, a1 = 2D0, b0 = K
 state and output equation:
x1 (t )  x 2 (t )
x 2 (t )   0 x1 (t )  2 D 0 x 2 (t )  u (t )
2

y (t )  K x1 (t )
 block diagram structure:

u(t) K
y(t)

2D0
2
0

© Prof. Dr.-Ing. Peter Firsching;Prof. Dr. László Juhász page 43


Advanced Modeling and Simulation - Mathematical Models of Physical Systems
Last Update: 15/03/2017 17:18:00

c)  state and output equation:


 x1 (t )   0 1 0   x1 (t )  0
 x (t )   0 0 1   x (t )  0 u (t )
 2    2   
 x 3 (t )   a 0  a 2  a 2   x3 (t )  1
 x1 (t ) 
y (t )  b0 b1 0  x 2 (t )
 x3 (t ) 
 block diagram structure:

b1

b0
u(t) y(t)

-a3

-a2

-a1

Question 31:

 observable canonical form:

 bm 
b 
  a n 1 1 0  0 
 x1  
m 1 
 x1   
 x   a n  2 0 1  0  x    
 2          2   b  u
     0 
    a1 0 0  1     0 
 x n    a 0 0 0 0   xn    
 0  
 0 
 
 x1 
x 
 
y  1 0  0  2

 
 xn 

© Prof. Dr.-Ing. Peter Firsching;Prof. Dr. László Juhász page 44


Advanced Modeling and Simulation - Mathematical Models of Physical Systems
Last Update: 15/03/2017 17:18:00

 diagonal form:

 x1   s1 0  0   x1   b1 * 
 x   0  s  0   x  b *
 2   2   2   2  u
         
      
 xn   0 0   sn   xn  bn *
 x1 
x 
y  c1 c2  cn   2 

 
 xn 

Question 32:

a) y (k )  a0 y (k  1)  b1 u (k )  b0 u (k  1)
b) y (k )  y (k  1)  6 y (k  2)  5u (k  1)

© Prof. Dr.-Ing. Peter Firsching;Prof. Dr. László Juhász page 45


Advanced Modeling and Simulation - Mathematical Models of Physical Systems
Last Update: 15/03/2017 17:18:00

Addendum: Laplace Transform – A short Introduction


A good introduction to the Laplace Transform is given in [DOB].
[WOL], appendix F, describes the way how to solve differential equations using the
Laplace transform.

Definition of the Laplace Transform / inverse Laplace Transfom


Given a function f(t) with f(t) = 0, t< 0. The Laplace Transform for f(t) is defined as:

F ( s)   f (t )e  st dt , s =  + j
0
The Laplace Integrale converge when  < 0.

The inverse Laplace Transform is written as:


  j
1
f (t )   F ( s)e st ds
2j   j

Ways of Representing the Laplace Transform

Laplace Transform Inverse Laplace Transform


F(s) = L{f(t)} f(t) = L-1{F(s)}
f(t) o--- F(s) F(s) ---o f(t)

Some Properties of the Laplace Transform

 Linearity of the LaplaceTransform


Given two functions f1(t) and f2(t) and two complex numbers a1, a2  C. In
the whole area of convergence of the Laplace transform is

L{a1 f1 (t )  a 2 f 2 (t )}  a1 L{ f1 (t )}  a 2 L{ f 2 (t )}

 Calculation of the derivative in time domain


Given a function f(t) with f(t) = 0, t< 0. The time derivative of the function
can be calculated for t  0. The Laplace Transform of the time derivative of
f(t) is
d
L{ f (t )}  s F ( s)
dt

© Prof. Dr.-Ing. Peter Firsching;Prof. Dr. László Juhász page 46


Advanced Modeling and Simulation - Mathematical Models of Physical Systems
Last Update: 15/03/2017 17:18:00

 Final value theorems


lim f (t )  lim s F ( s )
t 0 s 

lim f (t )  lim s F ( s )
t  s 0

Table of some important Laplace Transforms

F(s) f(t), t > 0 Name


1 0(t) unit pulse
1 -1(t) step
s
1 t -1(t) = -2(t) ramp
s2
1 e-at real pole
sa
1 1 1 step response first order
(1  e  at ) system
s sa a
1 1 1 two real poles
; ab (e  at  e bt )
sa sb ba
1 t e-at double real pole
( s  a) 2
02 e  D 0t (cos( 0 1  D 2 t )  step response second order
system
( s 2  2 D 0 s   0 ) s
2
D
sin( 0 1  D 2 t )
1 D 2

0<= D < 1

© Prof. Dr.-Ing. Peter Firsching;Prof. Dr. László Juhász page 47


Advanced Modeling and Simulation - Mathematical Models of Physical Systems
Last Update: 15/03/2017 17:18:00

Solution of linear time invariant differential equations using the Laplace


Transform
Step 1:
Given the model of a LTI system as differential equation:

y ( n ) (t )  a n1 y ( n1) (t )   a1 y (1) (t )  a 0 y (t ) 


bn u ( n) (t )  bn1 u ( n1) (t )    b1u (1) (t )  b0 u (t )
( ) d
The index () means: x (t )  x(t )
dt
d
Special case: x
(1)
(t )  x(t )  x (t )
dt

Step 2:
Using the properties of the Laplace Transform (linearity and calculation of the
derivative in time domain) the differential equation can be transformed to the
Laplace space.

s nY (s )  a n1s n1Y (s )    a1 sY (s )  a 0Y (s ) 
bn s nU (s )  bn1 s n1U (s )    b1 sU (s )  b0U (s )

This equation can be solved for Y(s) / U(s) = F(s). Thus we get the transfer
function
Y ( s ) bn s n  bn 1 s n 1  ....  b1 s  b0
F ( s)   n
U (s) s  a n 1 s n 1  ...  a1 s  a 0

Step 3:
Now the system reaction to a specific input signal can be calculated as
follows:
Y (s)  U ( s) * F (s)

The system reaction in time domain can be determined by transforming Y(s)


with the inverse Laplace transform

y (t )  L1{Y ( s )}  L1{U ( s) F ( s )}

The calculation of y(t) is usually based on the partial fraction expansion of


Y(s) and the inverse Laplace transform using transformation tables (see above
or addendum in Ch. 4).

© Prof. Dr.-Ing. Peter Firsching;Prof. Dr. László Juhász page 48


Advanced Modeling and Simulation - Mathematical Models of Physical Systems
Last Update: 15/03/2017 17:18:00

Example - step response function of an electrical RC network


Given a RC network (see also 1.2.1) which can be connected to an electric source by
a switch. When t = t0 the switch will be closed. Calculate the behaviour of the output
voltage u2(t).

1
Potential law: i1 (t )  (u1 (t )  u 2 (t ))
R t = t0
i1(t)
 1
Storage Law: u 2 (t )  i2 (t ) i2(t)
C u1(t)
R
u2(t)
US C
Balance Law: i1 (t )  i2 (t )  0

Result:

RC u 2 (t )  u 2 (t )  u1 (t )

The transfer function of the system can be written as:


U 2 ( s) 1
sRCU 2 ( s)  U 2 ( s)  U 1 ( s)  F ( s)  
U 1 ( s) 1  sRC

The input of the system can be interpreted as a step signal which occurs at t = t0.
Thus in the Laplace domain the input signal is:
1
U 1 ( s )  L{U S  1 (t )}  U S
s
Now the system output can be calculated as:
 1 1  1  1 
y (t )  L1 U S  US L  
 s 1  sRC   s (1  sRC ) 

The solution of the inverse Laplace transform can be found in the table given on
page 40.

y (t )  U S (1  e  t / T ) T  RC

© Prof. Dr.-Ing. Peter Firsching;Prof. Dr. László Juhász page 49

You might also like