University of Tripoli

Faculty of Engineering
Department of Mechanical Engineering

Automatic Control I ME330/450

Fall 2013

Part II
Linearization, Block Diagrams, Signal Flow
Graph and State Space

Azeddien Kinsheel
1
Linearization of Nonlinear Mathematical
Models
⚫ A system is nonlinear if the principle of
superposition does not apply.
⚫ Although many physical relationships are often
represented by linear equations, in most cases
actual relationship are not quite linear. In fact, a
careful study of physical systems reveal that
even so-called “linear systems” are rarely linear
only in a limited operational range.

2
Linearization of Nonlinear Mathematical
Models
⚫ In control engineering a normal operation of the
system may be around an equilibrium point , and the
signal may be considered small signal around the
equilibrium. (with exceptions to many cases).
Example:
The differential equation of the
pendulum system can be
obtained assuming equilibrium, T(t)
thus: θ

Ml  + Mgl sin( ) = T (t )

2 m
Mg
0
3
Linearization of Nonlinear Mathematical
Models

⚫ This is a nonlinear equation, however for small

change of θ around zero position, we can assume:
sin(θ)=θ

Ml 2(t ) + Mgl (t ) = T (t )

T(t)
θ
m

0
4
Mg
Linearization of Nonlinear Mathematical
Models

Linearized vs Non Linearized sin function

1

0.8 L
0.6
NL
0.4

0.2
Sin(tita)

-0.2

-0.4

-0.6

-0.8

-1
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2
Tita in rad

5
Linearization of Nonlinear Mathematical
Models
Linear approximation using Taylor series expansion
method:
Consider a system whose input is x(t) and output is y(t).
The relationship between y(t) and x(t) is given by:
Y=f(x)
If the normal operation condition corresponds to x0,y0 then
using Taylor series we get
y = f ( x)
df ( x ) 1 d 2 f ( x)
y = f ( x0 ) + ( x − x0 ) + 2
( x − x 0 ) 2
+ ....
dx x = x0 2! dt x = x
0
6
Linearization of Nonlinear mathematical
models

Where the derivatives df/dt, d2f/dt2 …….are evaluated at

x=x0
If the variation x-x0 is small, we may neglect the higher-
order terms in x-x0 then we can write
y = y0 + k ( x − x0 )
y0 = f ( x0 )
df
k=
dx x = x0

y − y0 = k ( x − x0 ) OR y = kx
Which indicates that y-y0 is proportional to x-x0
7
Linearization of Nonlinear mathematical
models
Example:
Linearize the following equation using Taylor series
expansion
y=x2 about x0=5 100

80

Solution:
y = f ( x) = x 2
60

df 1 d2 f y 40

y = f ( x0 ) + ( x − x0 ) + 2
( x − x 0 ) 2 + ....
dx 2! dt 20

df
f ( x0 ) = (5) 2 , = 10 and neglecting hiegher terms
dx x =5 0

we get : -20

y = 25 + 10( x − 5)
-40

( y − 25) = 10( x − 5)
0 1 2 3 4 5 6 7 8 9 10

x
y = 10 x
8
Transfer Function and Block Diagrams
⚫ In classical control systems control systems are often
represented by transfer functions and block diagrams
⚫ Transfer function: the transfer function of a linear ,
time-invariant differential equation system (LTI) is
defined as the ratio of the Laplace transform of the output
to the Laplace transform of the input, under the
assumption that all initial conditions are zeros.

9
Transfer Function and Block Diagrams
⚫ The transfer function can be represented mathematically as

C ( s ) bm S m + bm −1S m −1 + ... + b1S + b0 ( s − z1 )( s − z 2 )..( s − z m )

G (s) = = =
R( s ) an S n + an −1S n −1 + ... + a1S + a0 ( s − p1 )( s − p2 )...( s − pn )
where n  m
Or graphically

R(s) G(s) C(s)

10
Transfer Function
⚫ The application of the concept of transfer function is
limited to Linear Time-Invariant systems (LTI)
⚫ The transfer function is the property of the system
independent of the input or driving function.
⚫ The transfer function includes the units necessary to
relate the input to the output
⚫ The transfer function can be used to study the output of
a component, subsystem or a system under various
types of input.
⚫ The physical limitation of the system is not described in
the transfer function.

11
Transfer Function
⚫ System Order: is the highest power of S in the denominator
(n)
⚫ System Zeros: The roots of the nominator polynomial or the
values of s at the nominator that will give G(s)=0.
⚫ System Poles: The roots of the denominator polynomial or
the values of S at the denominator that will give G(s)=∞
⚫ System Type : is the highest order of s in the denominator
with zero root, or the number of zero poles.
⚫ Characteristic Equation of the system: is the denominator
polynomial when set to zero, because it determines the time
domain response .

12
Transfer Function

Example:
s 2 + 6s + 5 ( s + 5)( s + 1)
System order=4th G (s) = 4 = 2
s + 5s + 6 s
3 2
s ( s + 2)( s + 3)
Zeros: -5,-1
Poles: 0,0,-2,-3
Type=2
Characteristic equation: s + 5s + 6 s = 0
4 3 2

S-plane Im

O | X X O X
Re
-5 -4 -3 -2 -1 0
13
Reduction of Block Diagrams
Block diagram is the graphical representation of
dynamic systems and consists of unidirectional,
operational blocks that represent the transfer
function of the variable of interest.
There are three main components in block diagrams
- Gain Block R(s) G(s) C(s)

X2(s) +
- Summing Point X1(s)+X2(s)
+
X1(s)
- Pickoff point X(s)
G(s)

X(s) 14
 Block Diagram Reduction
⚫Reduction Rules
1-Combining block in cascade

X(s) X2(s) X1(s) X2(s)

x1(s) G1 G2 = G1G2

2-Moving summing point behind block

X2(s) + + X3(s)
G X3(s)
= X2(s) G
+ +
X1(s) X1(s)
G

15
 Block Diagram Reduction
3-Moving a pickoff point ahead of a block
X1(s) X2(s) X1(s) X2(s)
G
= G

X2(s) X2(s)
G

4-Moving a pickoff point behind a block

X1(s) X2(s) X1(s) X2(s)
G
= G
X1(s) X1(s)
1/G

16
 Block Diagram Reduction
5-Moving summing point ahead of a block
+ X1(s) + X3(s)
X1(s)
G
X3(s)
= G
+ +
X2(s) X2(s)
1/G
6-Eliminating the feedback loop

R(s) + C(s) R(s) C(s)

G
G
=
- 1 + GH

17
 Block Diagram Reduction
Example: Obtain the overall transfer function of
the following systems using block diagram
reduction.

H2
R(s) - C(s)
+ + +
G1 G2 G3 G4
+
-
H1

H3

18
 Block Diagram Reduction
Solution

C(s)
R(s) G1G2G3G4
1 − G3G4 H1 + G2G3 H 2 + G1G2G3G4 H 3

19
Signal Flow Graph
⚫ A Signal Flow Graph ”SFG” may be regarded as a
simplified version of a block diagram. The SFG
was introduced by S. J. Mason (53) for the cause-
effect representation of linear systems that are
modeled by algebraic equations. Beside the
difference in the physical appearance of the SFG
and BD, the signal-flow is constrained by more
rigid mathematical rules. Where BD notation is
more liberal.

20
Signal Flow Graph
⚫ A SFG may be defined as a graphical means of
prototyping the input-output relationship among the
variables of a set of linear algebraic equations.
⚫ For complicate system with complex
interrelationship BD reduction quit. difficult to
complete.
⚫ The advantage of SFG is the availability of SFG
Gain Formula to obtain the relation of the system
variables without reduction procedure

21
Signal Flow Graph
⚫SFG components
SFG consisting of R(s) G(s) C(s)
Nodes representing
signals or “system System
Input Output
variables”. Signal
Or
Signal
Gain
Nodes are connected
by branches that R(s) G(s) C(s)
representing
systems or ”gains”
node Branch
22
Signal Flow Graph
⚫ The node variable equals
to the summation of all
signals entering the node

V(s)=R1(s)G1(s)+R2(s)(-G2(s))+R3(s)G3(s)

C1(s)= V(s)G4(s)

C2(s)= V(s)G5(s)
C3(s)= V(s)(-G6(s))
23
Signal Flow Graph

⚫ Ri(s) nodes are called

input nodes or sources
⚫ Ci(s) are called output
nodes or sinks
⚫ V(s) is called mixed
node because it has
input and output
branches.
⚫ Forward path is a path
from an input node to an
output node that does
not cross any node more
than ones 24
Signal Flow Graph
⚫ A path in SFG is a
branch or a continuous
sequence of branches
that can be traversed Single path
from one node (signal) to
another node (signal)
⚫ A loop in SFG is a Multi
closed path that Paths
originates and terminates
on the same node. And
no node is met twice
along the path.
Single loop
25
Signal Flow Graph
⚫ The Gain in a path or
loop is the product of all
of the branch gains
(transmittance) in the Gain =G1(s)G2(s)G3(s)
path or the loop
⚫ Two loops are said to be G7
Non-touching if they do H 1
G1 Y5
not have a common G2 G4 G6
R(s) G5
node. G3 Y
Y1 Y2 3 H2 Y4 C(s)
⚫ Two loops are touching
if they share one or H3
more common nodes

26
Signal Flow Graph
⚫ The linear relationship Tk between
the input R(s) and the output C(s)
in signal flow graphs are obtained
using the Gain Formula:

T=
 k
Pk  k

Where:
Pk= Kth path from the input to the output.
Δ=Determinant of the graph.
Δk=cofactor of the path Pk.
And, the summation is taken over all possible k paths from
R(s)to C(s). 27
Signal Flow Graph

The cofactor Δk is equal to the determinant Δ with all the loops

touching the kth path substituted to zero.
And , the determinant Δ is:
Δ =1-(Sum of all different loops)

+(sum of the gain product of all combination of two non-touching loops)

-(sum of the gain product of all combination of three non-touching loops)
+….
N M ,Q
 = 1 −  Ln + L m Lq −  Lr Ls Lt +.....,
n =1 m =1, q =1

28
Signal Flow Graph
⚫ Properties of SGF
1- Branches are Unidirectional
2-Nodes add all the incoming branches and
transmits this sum to all the outgoing branches
3-A mixed node could be treated as output node
by adding a similar node with unity branch.
4-The same system may be represented by
different SFGs.

29
Signal Flow Graph
⚫ Example:
Use gain formula to obtain the transfer function
C(s)/R(s) of the system shown in the Figure

G6 G7

G1 Y5
G2 G4 1
R(s) G5
G
Y1 Y2 3 Y3 -H1Y4 C(s)

-H2

30
Signal Flow Graph
⚫ Solution:
1- Paths
P1=G1G2G3G4G5
P2=G1G6G4G5
P3=G1G2G7
G6 G7
2-Loops Y5
G1 G2
L1=-G4H1 G4 1
R(s) G5
L2=-G2G7H2 G
Y1 Y2 3 Y3 -H1Y4 C(s)
L3=-G6G4G5H2
L4=-G2G3G4G5H2 -H2

31
Signal Flow Graph
3- The determinant Δ
Non-touching loops L1,L2
Δ=1-(L1+L2+L3+L4)+(L1L2)
=1+G4H1+G2G7H2+G6G4G5H2+G2G3G4G5H2+G4H1G2G7H2

4-Cofactors Δi

Δ1=1 all loops touching path P1

Δ2=1 all loops touching path P2
Δ3=1-L1 all loops touching path P3 except L1
5- Finally applying the Gain Formula C (s)
=
 k
Pk  k
R( s) 
32
Signal Flow Graph

Substituting in the gain

Formula

C ( s ) P11 + P2  2 + P3  3
=
R( s) 

C (s) G1G 2 G 3G 4 G 5 + G1G 6 G 4 G 5 + G1G 2 G 7 (1 + G4 H1 )

=
R( s ) 1 + G4 H1 + G2G7 H 2 + G6G4G5 H 2 + G2G3G4G5 H 2 + G4 H1G2G7 H 2

33
Modeling in State Space
⚫ Modern control Versus Classical (Conventional)
control theory
Classical control theory is associated more with single
input signal output systems (SISO), Time-Invariant
Linear control systems using Transfer function and
Block diagram methods.
Modern Control theory in contrast is applicable to
Multi-Input Multi-Output (MIMO), Linear or nonlinear,
Time-Invariant or Time variant , using mostly signal
flow graph and state space representations.
34
Modeling in State Space
⚫ State:
The state of a dynamic system is the smallest set of
variables (called state variables) such that the
knowledge of theses variables at t=t0 , together of
the knowledge of the input for t≥t0. completely
determines the behavior of the system for any time t
≥ t0.
State variable: are the set of variables describing the
state of dynamic systems
State variables can be grouped together to form the
State Vector.
35
Modeling in State Space
⚫ State space : the n-dimensional space whose
coordinate axes consists of x1 axis, x2 axis, …xn
axis. Where x1, x2,..xi are the state variables, is
called State Space.
⚫ State Space equation: is the equation relating input
variables u , state variables x and output variables
y.

u1(t) Dynamic System y1(t)

State
u2(t) x1,x2,x.. y2(t)

36
Modeling in State Space

⚫ In state equation the state of the system is described by

a set of first order differential equations in terms of the
state variables x1,x2,x3..xn
⚫ The first order differential equations for n state variables
and m inputs can be written in general form as:
x1 = a11 x1 + a12 x2 + ..a1n xn + b11u1 + ...b1mum
x2 = a21 x1 + a22 x2 + ..a2 n xn + b21u1 + ...b2 mum

xn = an1 x1 + an 2 x2 + ..ann xn + bn1u1 + ...bnmum

Where dx
x =
dt
37
Modeling in State Space
⚫ This set of simultaneous differential equations
can be written in matrix form as follows:
 x1   a11 a12  a1n   x1 
      b11  b1m   u1 
 a 2 n   x2  
+        
d  x2  a21 a22
=
dt           
      bn1  bnm  um 
 xn  an1 an 2  ann   xn 

Where X is the state  x1 

vector , x1,x2..xn are x 
the state variables
X = 2 

 
 xn  38
Modeling in State Space
⚫ The state Differential equation is normally
written in a compact form as:

x = Ax + Bu
The above equation also called State Equation relates the
rate of change of the state of the dynamic system to the
state of the system and the input signals.

39
Modeling in State Space

However the output of the system can be related to the

state of the system and the input signals by the Output
Equation

y = Cx + Du

Where y is the set of output signals expressed

in column vector form.
These two equations are called state –variable
representation or state-space representation of a
system
x = Ax + Bu

y = Cx + Du

40
Modeling in State Space
⚫ Example: write the
state-variable equation
of the system shown in
the figure.

Inputs:
u=f(t)
Outputs
y=x(t)
The differential d 2 x(t ) dx(t )
equation of the M 2
+ f v + Kx (t ) = f (t )
d t dt
system is
41
Modeling in State Space

Let x1=x(t)
X2=dx(t)/dt
Then, the differential equation of the system can
be written in terms of the state variables x1 and x2
as: x = x
1 2

fv K 1
x2 = − x2 − x1 + u (t )
M M M
y = x1
rewritting in Matrix - Vector Form we get :
 x1   0 1  x   0 
 x  =  − K − f v   1  +  1 u (t )
 2   M M   2   M 
x
and
x 
y = 1 0 1  + 0u
 x2 
42
Modeling in State Space

The previous equations coincide with the standard

form of the state-variable equations:

x = Ax + Bu
y = Cx + Du
Where :
 0 1 
A =  − K − fv 
 M M 
0
B= 1 
 M 
C = 1 0
D = 0
43
Modeling in State Space

⚫ Obtaining Transfer Function from State Space

equations:
Let the transfer function of a SISO system given by
Y ( s)
G( s) =
U ( s)
and the state space equation of the system is
x = Ax + Bu
y = Cx + Du
Then by applying LT. to the state equations we get
SX(s) - x(0) = AX(s) + BU(s)
Y(s) = CX(s) + DU(s)

44
Modeling in State Space

Since the transfer function of a system is the ratio of the Lt

of the output to the Lt of the input with all initial
conditions are zeros then
X(0)=0
Y ( s)
G( s) =
U ( s) I is the unity matrix
SX(s) = AX(s) + BU(s) and its size is the
same as A
( SI − A) X ( s ) = BU ( s )
or
X ( s ) = ( SI − A) −1 BU ( s )
45
Modeling in State Space

Substituting by the value of X(s) in the LT of the

output equation we get
Y ( s ) = C[( SI − A) −1 BU ( s )] + DU ( s )
Y ( s ) = (C ( SI − A) −1 B + D)U ( s )
Y(s)
since G(s) = , Then
U(s)
Y(s)
G (s) = = C ( SI − A) −1 B + D
U(s)

The last equation gives an expression to calculate the transfer

function of a system using SS equation coefficients A,B,C and D.

46
Modeling in State Space

⚫ The characteristics equation of the transfer function G(s)

can be obtained from the determinant of the matrix (SI-
A)-1
Where G(s) can be written as :
Q( s)
G (s) =
SI − A
Transfer Matrix:
The same equation of transfer function can be used to obtain the
transfer functions of r inputs and m outputs system. however the
size of G(s) the transfer matrix is mxr. Each element Gi,j(s) of
G(s) will represent the transfer function of Yi(s)/Uj(s)

47
Modeling in State Space

⚫ State Transition equation:

After obtaining the state space equation , we may
need to find the solution for that equation for
further time domain analysis. For this purpose we
may use classical methods or LT method.
Using LT method we can solve the SS equation as
follows:

48
Modeling in State Space

[ x ] = [ Ax + Bu ]
SX ( s ) − x(0) = AX ( s ) + BU ( s )
X ( s ) = ( SI − A) −1 x(0) + ( SI − A) −1 BU ( s )
and
x(t ) =  −1[( SI − A) −1 ]x(0) +  −1[( SI − A) −1 BU ( s )]
let
 (t ) =  −1[( SI − A) −1 ] = e At
where  (t)is called the transition matrix
then
t
x(t) =  (t)x(0) + B   (t -  )u ( )d
0

49
Modeling in State Space

⚫ State Diagram:
State diagram is an extension to SFG to portray state
equations and differential equations
To construct state diagram from differential equations
1-The DF is arranged in the form:
d n y (t ) d n −1 y (t ) dy (t )
= − a − ..... − a − a1 y (t ) + r (t )
dt n −1
n 2
dt n dt

2- Arrange the nodes R(s), sny(s),sn-1y(s)…. From Left to right

R snY Sn-1Y SY Y

50
Modeling in State Space
3- The nodes are connected by branches

R snY Sn-1Y SY Y
-an
1
-a2

-a1

51
Modeling in State Space
4-Add the integrators and initial conditions

y(n-1)(t0)/s y(1)(t0)/s y(t0)/s

Sn-1Y
S-1 Y Y
R S-1
-an 1
1 xn SY x x1
2
snY -a2

-a1

52
Modeling in State Space
⚫ Example:
Draw the space diagram of the following
differential equation

d 2 y (t ) dy (t )
2
+3 + 2 y (t ) = r (t )
dt dt

53
Modeling in State Space

⚫Solution

y(1)(0)/s y(0)/s

s2Y SY
R S-1 S-1 Y Y
-3
1 x2 x1
1
-2

54
Modeling in State Space
⚫ From state Diagram to state equation and output
equation
1- Delete the initial states and branches with
integrators
2-for SS equation regard states represent
derivatives as output and
3- regard the inputs and the state variables as an
inputs
4- Apply gain formula to each combination
5- substitute in the SS equation coefficients A, B,
C and D
55