Sie sind auf Seite 1von 43

1

INTRODUCTION TO STATE SPACE ANALYSIS

9.1 Introduction

The analysis and performance of control systems as given in previous chapters is

based on transfer function and graphical approach such as Bode, Nyquist and root locus.

The mathematical model of the control system is being obtained in frequency domain

(s- domain). The basis of transfer function is:

 Applicable for linear time invariant system only.

 Applicable only for Single Input Single Output (SISO) systems.

 Initial conditions in system equation are considered as zero.

 The analysis of system output for a specified input is obtained. No information

regarding initial state of the system is available.

 It is complex to use computer.

 It is not possible to design the system to its operational performance.

The modern approach known as state space analysis is preferred to transfer

function approach, which considers analysis in time domain and over comes the

limitations of transfer function approach. The state space model of a control system takes

into account the following advantages

 Applicable for linear time invariant / variant system.

 Applicable for SISO and Multiple Input and Multiple Output systems (MIMO).

 The initial conditions pertaining to the system are taken into account.

 The state space model gives complete description of the system

 The analysis is carried out in time domain and easy amenable to solve through

digital computer.

Lecture Notes by N.Selvaganesan, IIST (January – May 2018)


2

The modern state space approach need not replace classical transfer function approach.

The transfer function approach gives a physical insight into the system for preliminary

design of control system through simple mathematical model.

9.2 Definitions

1. State variables

The smallest of variables which determine the state of a dynamic system are

called state variables. It is not necessary that the state variable be physically

measurable or observable quantities.

2. State

The state of the system is the smallest set of state variables such that the

knowledge of variable at t = t0 (initial condition) together with the inputs completely

determine the behaviour of the system for any time t > t0.

3. State vector

If „n‟ state variables are necessary to determine the behaviour of a given system,

the variables are considered as n components of a vector called state vector.

4. State space

The „n‟ dimensional state variables are elements of „n‟ dimensional space called

state space.

9.3 State space representation

The state model of linear time invariant system can be written in the matrix form as


X (t) = A X(t) + B U(t)  State Equation
(n×1) (n×n) (n×1) (n×m) (m×1)

Lecture Notes by N.Selvaganesan, IIST (January – May 2018)


3

Y(t) = C X(t) + D U(t)


 Output equation
(p×1) (p×n) (n×1) (p×m) (m×1)

where, X(t) is a state vector

A is state matrix of order n×n

B is the input matrix of order n×m

C is the output matrix of order p×n

D is the transmission matrix of order p×m

9.3.1 State space representation for Physical systems

This section gives an idea of formulating state model for a given physical system

using physical variables.

Problem 9.1

Formulate the state equation for the given electrical network shown in Fig 9.1.
R L

E I V C

Fig 9.1
Solution

State variables are chosen as (no of charging elements equal to no of state variables)

x1 = I and x 2 = V

Lecture Notes by N.Selvaganesan, IIST (January – May 2018)


4

Apply Kirchoff‟s law and write the differential equation

dI dI E R V
E = RI + L +V  = - I-
dt dt L L L

Rewritting the above equation

  E R V
I=x= - I- (a)
L L L

dV  1
But I = C  V= I (b)
dt C

From the equation (a) and (b) state equation in matrix form is

 R 1   
      - -   
 I    x1  =  L L   x1  +  1  V
     1    
 V   x 2    x
0  2  
L
 C  0 

and the output equation is

x 
Y = V = x 2  Y = 0 1  1 
x2 

Problem 9.2

Obtain the state model for the system shown in Fig 9.2

R V2 R V1

U I1 I2 C
C

Fig 9.2

Solution

Apply Kirchoff‟s current law (nodal method) and write the differential equation

Lecture Notes by N.Selvaganesan, IIST (January – May 2018)


5

V2 - U dV V - V
At Node V2  + C 2 + 2 1 = 0 and (a)
R dt R

V1 - V2 dV
At Node V1  +C 1 = 0 (b)
R dt

Rearranging the equation (a) and (b)

dV2 U - V2 V2 - V1  U - 2V2 + V1
C = -  V2 =
dt R R RC
 V1 - V2 V V
V1 = - =- 1 + 2
RC RC RC

In matrix form

 1 1  V 
   -  1 0 
 V1  =  RC RC    +  1  U
   1 2    
 V 2   -  V2   RC 
 RC RC 

and the output equation is

Y = V1 = x1

x 
Y = 1 0   1 
x2 

Problem 9.3

Obtain the state model for the system shown in Fig 9.3

R1 L1 L2

I1 VC I2 R2
V C

Fig 9.3
Lecture Notes by N.Selvaganesan, IIST (January – May 2018)
6

Solution

State variables are chosen as (no of charging elements = no of state variables = 3)

x1  t  = I1  t 

x 2  t  = I2  t 

x 3  t  = Vc  t 

Apply Kirchoff‟s law (loop method) and write the differential equation

dI1
V  t  = R1I1 (t) + L1 + Vc  t  (a)
dt

dI2
R 2 I 2  t  + L2 = Vc  t  (b)
dt

dVc
C = I1  t  - I2  t  (c)
dt

From (a), (b)and(c)

 R1 1 1
x1  t  = - x1  t  - x 3  t  + V  t 
L1 L1 L1

 R2 1
x2  t  = - x 2  t  + x3  t 
L2 L2

 1
x3  t  =  x  t  - x 2  t 
C 1 

In matrix form, state model can be written as

Lecture Notes by N.Selvaganesan, IIST (January – May 2018)


7

    R1 
 x1  - L1 0 - L1   x1   L1 
1 1

   =  0 - R2    
x2   L2   x 2  +  0  V  t 
1
L2
     x  0 
 1 0  3   
1
 x 3   C C 
 x1 
 
y  t  = 0 R 2 0  x 2 
 x 3 

Problem 9.4

Formulate the state equation for the given mechanical system given in Fig 9.4. Choosing

displacements y1 and y2 of the mass M1 and M2 are state variables as well as output.

K2
B y1 y2
K1

M1 y1 K1 M2 F
M1
K2 B

M2 y2
Fig 9.4 Mechanical Network
Fig 9.5

Solution

By applying Newton‟s law of motion to the mechanical network (shown in Fig 9.5), the force

equation can be written as

Equilibrium equations are written below

d 2 y1 dy
At node (y1)  M1 +B 1 +K1 y1 +K 2 (y1 -y 2 )=0 (1)
dt 2 dt

Lecture Notes by N.Selvaganesan, IIST (January – May 2018)


8

d2z2
At node (y2)  M 2  K 2 ( z 2  z1 )  f ( t ) (2)
dt 2

The state variables are selected as follows

y1 = x1 y2 = x 3

y1 = x1 = x 2 y2 = x 3 = x 4

.. . .. .
y1 = x 2 y2 = x4
. .. . ..
Substitute above values of z1 , z1 , z1 and z 2 , z 2 , z 2 in equations (1) an (2) to obtain

following set of equations:

x1 = x 2

M1 x 2 +Bx 2 +K1x 3 +K 2 (x1 -x 3 )=0

x3 = x 4
.
M1 x 4 +K 2 (x 3 -x1 ) = f(t)
Select displacements of mass M1 and M2 as output variables

c1 = y1 and c2 = y 2

Based on above set of equations the state model is developed as below

.
   
 x1   0 1 0 0  x 
    1  0 
. 0    0 
(K1 +K 2 ) B K2
  - -
x2  =  M1 M1 M1  x2   
  +  0  f(t)
 
. 0 0 0 1    
 x3    x3   1 
   K2 K2 
0    M 2 
   M1
. 0
 x 4 
M2   x 
 4

 c1  1 0  x 1 
and  c  = 0 1   x 
 2    2

Lecture Notes by N.Selvaganesan, IIST (January – May 2018)


9

9.3.2 Electro mechanical systems

DC motors used in servo systems can be operated almost maintenance free

(electronic commutation) called “Brushless” dc motor.

DC servo motor

The following are the special features of DC servo motor

 High torque-inertia ratio.

 Small time constant.

 Small power ratings are used in computer related equipments such as disc drives,

printers and tape drivers.

 Medium and large power ratings are used in robot systems and CNC machines.

 Some servo motor magnetic field is produced by separated excited /permanent

magnet.

 Normally armature controlled dc motor is used.[IF = constant]

 Speed controlled by varying armature voltage is called “Armature control dc

motor”.

 Speed is controlled by Field current is called “Field controlled dc motor”.

Disadvantages

 Requirement of constant armature current is a series disadvantage, because it is

more difficult to provide constant current than constant voltage.

 Time constant of the field controlled dc motor are generally large.

9.3.2.1 Armature control of dc servomotor

The Fig 9.6 shows the circuit model for the Armature control of dc servomotor.

Ia Ra Rf IF

Lf – MayI 2018)
Lecture Notes by N.Selvaganesan, IIST (January VF
La f
Va Ia
10

From the chapter 3, section 3.7.1, the transfer function of the armature control speed DC

motor is obtained as

θ s kT
=
Va  s  s  La Js 2 +  La B + R a J  s + R a B + k T k b 
 

The inductance La in Armature circuit is usually small and may be neglected

θ s  Km
= (a)
Va  s  s  Tms +1

where,

kT
km = = Motor gain constant
 R a b + kTkb 
RaJ
Tm = = Motor time constant
 R a B + kTk b 
State space representation

Rearranging the equation (a)

 1  km
θ+ θ= V
Tm Tm a

Defining state variables x1 and x 2

x1 = θ = y


x2 = θ

Lecture Notes by N.Selvaganesan, IIST (January – May 2018)


11

and hence

 
x1 = θ = x 2

  1 k
x2 = θ = - x 2 + m Va
Tm Tm

State space representation of the Dc armature control motor is given by

 x1  0 1   x1  0 
    
  = 0 1    +  K m  Va
-
 x 2   Tm   x 2   Tm 

Output equation is given by

y = θ = x1

x 
y = 1 0   1 
x2 

9.3.2.2 Field control of dc servomotor

The Fig 9.7 shows the circuit model for the Field control of dc servomotor.

Rf

Lf M Ia
E If Va

J 
Fig 9.7
B

Lecture Notes by N.Selvaganesan, IIST (January – May 2018)


12

From the chapter 3, section 3.7.2, the transfer function of the field control speed DC

motor is obtained as

θ s  Km
= (a)
E  s   τme τΒs + τmes2 + τ Bs2 + s 
3

where,

kT
Km = = Motor gain constant
R FF

LF
τB = = Time constant for field circuit
RF

J
τme = = Mechanical time constant
F

Rearranging the equation (a)

  
θ τme τΒ + θ  τme + τ B  + θ = E  s  K m


θ+θ
  τme + τB  + θ 1 k
= m E
τme τ Β τ me τ Β τ me τ Β

Defining state variables x1, x 2 and x3

x1 = θ

x2 = θ

x3 = θ
and hence

 
x1 = θ = x 2
 
x 2 = θ = x3
 
x3 = θ = -
 τme + τF  θ- 1  km
θ+ E
τ me τ F τ me τ F τ me τ F

Lecture Notes by N.Selvaganesan, IIST (January – May 2018)


13

State space representation of the Dc field control motor is given by

     x1   0 
 x1  0 1 0
   
      
 x 2  = 0 0 1   x 2  + 0 E
    τ me + τ F 
   
 x 3  0 -     km 
1
-
   τ me τ F τ me τ F   x   τ me τ F 
 3

Output equation is given by y = θ

 x1 
y = 1 0 0  x 2 
 x 3 

9.3.3 State modeling using phase variable

In this section an alternate state space representation of control systems using

phase variables as state variables is discussed. Phase variable are defined as those

particular state variable which are obtained from one of the system variable and its

derivatives. The phase variable state model is easily determined if the model is already

known in the differential equation / transfer function.

Consider a system (LTIV) whose differential equation is

 
yn + a1yn-1 + a 2 yn-2 +......+ a n-1 y+ a n y = b0u m + b1u m-1 +......+ b m-1 u+ b mu

For time invariant system a i and bi are constant. The transfer function for the above

differential equation is

y  s  b0sm + b1sm-1 +.....+ b m


= n
u s  s + a1s n-1 +.....+ a n at IC=0

where, m =Number of Zero‟s

n = Number of Poles

Lecture Notes by N.Selvaganesan, IIST (January – May 2018)


14

Case: 1(with poles only)

y s  b
Transfer function =  n (or)
u  s  s + a1s +.....+ a n
n-1

yn + a1yn-1 +.....+ a n y = bu

Choose the state variable as

x1 = y


x2 = y


x3 = y

:
:
x n = y n-1

and is reduced to a set of n first order differential equations given below


x1 = x 2


x 2 = x3

:
:

x n-1 = x n


x n = -a n x1 - a n-1x 2 -.....- a1x n + bu

The above equation result in the following state equations:

Lecture Notes by N.Selvaganesan, IIST (January – May 2018)


15

   0 1 0................0   x1  0 
 x1    
   0 0 1................0   x 2  0 
 x 2  :  :  : 
:  =   + u
  :  :  : 
 x  
 n-1  0 0 0...............1   x n-1  0 
   
   -a n - a n-1 - a n-2 ..........a1   x n   b 
xn  

Note

It has all ones in the upper half diagonal, its last row is comprised of the negative

of the co-efficient of the original differential equation and all other elements are zero.

This form of matrix „A‟ is known as “BUSH FORM (OR) COMPANION FORM”.

Output equation

y = x1

 x1 
 
x2 
y = 1 0.........0 : 
 
: 
x 
 n

Case: 2(with poles and zeros)

Y  s  b0s3 + b1s 2 + b 2s + b3
Let us consider the T.F (proper system m = n =3) T  s  = = 3
U s  s + a1s 2 + a 2s + a 3

Method: 1

Y  S X1  S Y  S 1 b0s3 + b1s 2 + b 2s + b3
= = 
U  S U  S X1  S s3 + a1s 2 + a 2s + a 3 1

X1  S 1   
=  x + a x+ a x+ a3 = u
U  S s3 + a1s 2 + a 2s + a 3 1 2

Lecture Notes by N.Selvaganesan, IIST (January – May 2018)


16

From the physical variable method, state equation can be written as

 
 x1  0 1 0   x1  0
      
 x 2  = 0 0 1   x 2  + 0  u
   -a - a - a   x  1 
x3   3 2 1   3   
 

Y  S
and = b0s3 + b1s 2 + b 2s + b3
X1  S

y = b0 x1 + b1 x1 + b 2 x1 + b3

Choose the state variable as

x1 = y


x2 = y


x3 = y

The above equation result in the following state equations:

 
x1 = y = x 2
 
x1 = y = x 3
   
x 1 = y = u - a 1 x1 - a 2 x 1 - a 3 x 1

and its output equation is

y = b0  u - a1x 3 - a 2 x 2 - a 3 x1  + b1x 3 + b 2 x 2 + b 3x
= x1  b3 - a 3b 0  + x 2  b 2 - a 2 b 0  + x 3  b1 - a 1b 0  + b 0u
 x1 
 
y =  b3 - a 3 b 0 b 2 - a 2 b 0 b1 - a1b 0   x 2  + b 0 u
 x 3 

In general, the state model can be written as:

Lecture Notes by N.Selvaganesan, IIST (January – May 2018)


17

 
 x1   0 1 0...........0   x1  0 
    
 x 2  0 0 1...........0   x 2  0 
:  = :  :  + :u
      
:  :  :  : 
   -a - a    1 
xn   n n-1 - a n-2 .......- a 1   x n 

 

 x1 
x 
y =  b n - a n b0  b n-1 - a n-1b0  ......  b1 - a1b0    2  + b0 u
: 
 
xn 

Method: 2

y  S 1
Let us consider the transfer function T  S = =  PK Δ K
u  S Δ K

where,

Δ = 1-Sum of gain product of non-touching loops + Sum of gain products of two non-

touching loops -…..

PK = Path gain of Kth forward path.

b 0s3 + b1s 2 + b 2s + b 3
Consider the transfer function T  s  =
s3 + a1s 2 + a 2s + a 3

Choose the state variable as x1, x2 and x3 (third order system). By rearranging the above

b1 b2 b3
s3 = y  s 
b0 + + +
equation T  s  =
2
s s
1-  - 1 - 2 2 - 3 3  u  s 
a a a
 s s s 

From the signal flow graph (given in Fig 9.10) the following state equation is written.

Lecture Notes by N.Selvaganesan, IIST (January – May 2018)


18


x1 = b1u + x 2 - a1  x1 + b 0 u 
= -a1x1 + x 2 +  b1 - a1b 0  u

x 2 = x 3 - a 2  x1 + b 0 u  + b 2 u
= -a 2 x1 + x 3 +  b 2 - a 2 b 0  u

x 3 = -a 3  b 0 u + x1  + b3u
= -a 3 x1 +  b3 - a 3b 0  u
b0

b1
b2

1 1 1
x3 x3 x2 x2 x1 s x1
s s
u 1 b3 1 y
-a1

-a2
Fig 9.10 -a3
(or)

 
 x1  -a1 1 0   x1   b1 - a1b0 
      
 x 2  = -a 2 0 1   x 2  + b 2 - a 2 b 0  u
   -a 0 0  x   b - a b 
x3   3  3  3 3 0 
 

and the output equation is

y = x1 + b 0 u

 x1 
y = 1 0 0  x 2  + b0 u
 x 3 

In general, the state model can be written as:

Lecture Notes by N.Selvaganesan, IIST (January – May 2018)


19

 
 x1  -a1 1 0.....0   x1   b1 - a1b0 
      
 x 2  -a 2 0 1.....0   x 2   b 2 - a 2 b0 
: =: : + : u
      
:  : :   : 
   -a 0 0.....0   x   b - a b 
xn   n  n  n n 0
 

 x1 
x 
y = 1 0.......0  2  + b 0 u
: 
 
xn 

Lecture Notes by N.Selvaganesan, IIST (January – May 2018)


20

9.3.4 State modeling by Canonical Variable (CV) / Normal Form

Consider the transfer function

y  s  b0s m + b1s m-1 +..... + b m


= n
u s s + a1s n-1 +..... + a n

CV represents of a system; the matrix A turns out to be a diagonal matrix. This form of

state model plays an important role on control theory

i) Assume the system poles are distinct values λ1 , λ 2 .......λ n

y s n
c
= T  s  = b0 +  i
u s  i=1 s - λ i

where ci = Residues of the poles at s = λi

The block diagram model for the transfer function is shown in Fig 9.11.

u x1 x1 y
1
+ C1 +
- s

λ1

x2 1 x2
+ C2
- s

λ2


xn 1 xn
+ Cn
- s

λ3

bo
Fig 9.11
Lecture Notes by N.Selvaganesan, IIST (January – May 2018)
21

From the block diagram the following state and output equation are written.


x1 = λ1x1 + u

x 2 = λ2x 2 + u

:
:

x n = λn x n + u

y =  c1x1 + c2 x 2 +...... + cn x n  + b0 u

The state model can be expressed in the vector matrix form as

 
 x1   λ1 0 0........0   x1  1
  
 x 2  0 λ 2 0........0   x 2  1
:  = :  :  + :  u
      
:  :  :  : 
   0 .................... λ   x  1
xn   n  n  
 

 x1 
x 
y =  c1 c2 ...... cn   2  + b0 u
: 
 
xn 

Note

It is observed that for the canonical model, the system matrix A is a diagonal

matrix with the poles of T(s) as its diagonal elements. It is also observed that the

elements of column vector B of the model are all unity and the elements of the row vector

C are the residues of the system poles.

An alternate canonical state model is given below

Lecture Notes by N.Selvaganesan, IIST (January – May 2018)


22

 
 x1   λ1 0 0........0   x1  c1 
     x  c   x1 
  
x 2 0 λ 2 0........0    
2 2 x 
: =:   :  +  :  u „; y = 1 1......1  2  + b 0 u
       : 
:  :    
: :  
   0 .................... λ   x  c  xn 
xn   n  n  n

 

ii) With repeated pole at λ1 with multiplicity ‘y’ and remaining poles at λ r+1....λ n are

distinct.

Transfer function in Partial Fractional Form

y s r
ci n
ci
T s  = = b0 +  + 
u s  i=1  s - λ i 
r-i+1
i=r+1 s - λ i
  
u xr 1 xr x2 1 x2 x1 x1
+ 1
+
- s - s - s

λr λ2 λ1

Cr C2 C1

+ y


x r 1 1 x r 1
+ Cr+1
- s

λr


xn 1 xn
+ Cn
- s

λn Fig 9.12
Lecture Notes by N.Selvaganesan, IIST (January – May 2018)
23

A block diagram model for this case is shown in above figure.


x = Jx + Bu
y = Cx + b0 u

where,

 λ1 1 0.....0 0 0 0..... 0 
0 λ 2 1.....0 0 0 0..... 0 

: : : : : : : : 
 
0 0 0.....λ r-1 1 0 0 0 
J = 0 0 0.....0 λr 0 0 0 
 
0 0 0.....0 0 λ r+1 ...... 0 
: : : : : : : : 
 
: : : : : : : : 
 
0 0 0.....0 0 0.........λ n 

B =  0 0 0 1 1.....1
T

C = c1 c2 .......cr 

Note

Matrix J is called Jordan canonical matrix and has the following properties:

 The diagonal elements of J are poles of T(s)

 All the elements below the principal diagonal are zero

 A certain number of unit elements are contained immediately to the right of

principal diagonal when the adjacent elements in the principal diagonal are equal

Problem 9.5

  
Consider the given equation defined by y + 6 y+11y+ 6y = 6u where „y‟ is output and

„u‟ is input of the system. Obtain the state space representation of the system and draw

the block diagram.


Lecture Notes by N.Selvaganesan, IIST (January – May 2018)
24

Solution

i) Physical variable form

Choose the state variable as

x1 = y

x2 = y

x3 = y

The above equation result in the following state equations:

 
x1 = y = x 2

 
x 2 = y = x3

    
x3 = y = 6u - 6 y-11 y- 6y  x 3 = 6u - 6x1 -11x 2 - 6x 3

State space representation of the system and its block diagram is given in Fig 9.13.

 
 x1  0 1 0   x1   0  x1 
      
 x 2  = 0 0 1   x 2  + 0 and y = x1 = 1 0 0  x 2 
   -6 -11 - 6  x  6  x 3 
x3    3  
 

 

x 2  x3 x1  x 2 x1 y
u x3 1 1 1
+
- s s s

-6

-11

-6
Fig 9.13

Lecture Notes by N.Selvaganesan, IIST (January – May 2018)


25

ii) Canonical variable form

y s 6 6
G s  = = 3 =
u  s  s + 6s +11s + 6  s +1 s + 2 s + 3 
2

A B C
= + +
 s +1  s + 2   s + 3
Taking partial fraction

6 = A  s + 2  s + 3 + B  s +1 s + 3 + C  s +1 s + 2 

Solving the above equation

A=3

B = -6

C=3

3 -6 3
y s  = u s  + u s  + u s 
s +1 s+2 s +3

= x1  s  + x 2  s  + x 3  s 

where,

3 -6 3
x1  s  = u s  , x 2 s  = u  s  and x 3  s  = u s 
s +1 s+2 s +3

Rearranging the equation

 
x1  s  + x1  s  = 3u  t   x1  s  = -x1  s  + 3u  t 


x 2  s  = -6u  s  - 2x 2  s 


x 3  s  = 3u  s  - sx 3

State space representation of the system and its block diagram is given in Fig 9.14

Lecture Notes by N.Selvaganesan, IIST (January – May 2018)


26

 
 x1  -1 0 0   x1   3 
  
x2  =  0 - 2 0   x 2  + -6 u  t 
   0 0 - 3   x 3  3 
x3  
 

 x1 
y(t) = 1 1 1  x 2 
 x 3 

Note

 Diagonal elements are nothing but roots of the characteristic equation.

 State variables are not unique. But transfer function is unique.


u x1 x1 y
1
+ 3 +
- s

-1

x2 1 x2
+ -6
- s

-2


x3 1 x3
+ 3
- s

-3
Fig 9.14

Lecture Notes by N.Selvaganesan, IIST (January – May 2018)


27

Problem 9.6

y s 10  s + 4 
A Feed back system has a closed loop transfer function = Construct
u  s  s  s +1 s + 3

different state model for this system and give block diagram representation for each state

model.

Solution

i) Using canonical variable

Applying partial fraction

10  s + 4  A B C
= + +
s  s +1 s + 3 s s +1 s + 3

10  s + 4  = A  s +1 s + 3 + B  s  s + 3 + C  s  s +1

Solving the above equation

A = 13.33, B = -15 and C = 1.67

y s 13.33 -15 1.67


= T s  = + +
u s  s s +1 s + 3

State space representation of the system is given by

0 0 0   x1  1
x = 0 -1 0   x 2  + 1 u

0 0 - 3  x 3  1

 x1 
y = 13.33 -15 1.67   x 2 
 x 3 

ii) Using Phase variable

y  s  x1  s  y  s  1
T s = =  = 3  10s + 40 
u  s  u  s  x1  s  s + 4s 2 + 3s

Lecture Notes by N.Selvaganesan, IIST (January – May 2018)


28

x1  s  1
= 3
u  s  s + 4s 2 + 3s

Choose the state variable as x1 , x 2 and x 3 .

The above equation result the following state equations


x1 = x 2


x 2 = x3

  
x 1 + 4 x1 + 3 x 1 + 0 x 1 = u 1

State equation representation of the system is given by

 
 x1   0 1 0   x 1   0 
      
 x 2  = 0 0 1   x 2  + 0 u
   0 - 3 - 4  x  1 
x3    3  
 

Output equation

y s
= 10s + 40
x1  s 

y  s  = 10sx1  s  + 40x1  s 
= 10x1 + 40x1
= 10x 2 + 40x1

x 
y =  40 10  1 
x2 

Problem 9.7

Draw a suitable signal flow graph and from that deduce a state model of the system

10s + 40
T s  =
s + 4s2 + 3s
3

Lecture Notes by N.Selvaganesan, IIST (January – May 2018)


29

Solution

Divide both numerator and denominator by s 3

10 + 40 s3 10 + 40
T s =
s2 2
s3
= s
1 + 4 s + 3 s2 1 - ( 4 s - 3 s2 )

A signal flow graph configuration which satisfies the above requirements is show in Fig
10
9.15.

1 1 1
x3 x3 x2 s x2 x1 s x1
s
1 y
u 1 40
-4

-3

Fig 9.15
From the flow graph, we have


x1 = x 2 - 4x1


x 2 = 10u + x 3 - 3x1


x 3 = 40u

and y = x1

State space representation of the system is given by

-4 1 0   x1  0 
x = -3 0 1   x 2  + 10  u

 0 0 0   x 3   40 

 x1 
y = 1 0 0  x 2 
 x 3 

Lecture Notes by N.Selvaganesan, IIST (January – May 2018)


30

Problem 9.8

s 2 + 2s + 3
A Feedback system characterized by closed loop T.F. T  s  = 3 . Draw the
s + 2s2 + 3s +1

suitable signal flow graph and therefore construct state model of the system.

Solution

Divide both numerator and denominator by s 3

+ 2 s 2 + 3 s3
1
T s =
s

1 - - 2 s - 3 s2 - 1s3 
1

1 1 1
x3 x3 x2 s x2 x1 s x1
s
1 y
u 1 3
-2

-3
Fig 9.16
-1

From the flow graph (in Fig 9.16), we have


x1 = u + x 2 - 2x1


x 2 = 2u - 3x1 + x 3


x 3 = 3u - x1

and y = x1

State space representation of the system is given by

Lecture Notes by N.Selvaganesan, IIST (January – May 2018)


31

-2 1 0   x1  1 
x = -3 0 1   x 2  +  2 u

-1 0 0  x 3  3 

 x1 
y = 1 0 0  x 2 
 x 3 

9.4 Determination of Transfer Function from State Model

Let the state model be given as


x = AX + BU (1)

y = CX + DU (2)

Take Laplace transform on (1) and (2)

sx  s  - x  0  = Ax  s  + Bu  s  (3)

y  s  = Cx  s  + Du  s  (4)

Rearranging the equation (3)

sI - A  x  s  = x  0  + Bu  s 

x  s  = sI - A   x  0  + Bu  s  
-1
(5)

Substitute (5) in (4)

y  s  = C sI - A x  0  + sI - A Bu  s  + Du s 
-1 -1
 

Under zero initial condition x  0  = 0;

y s
= C sI - A  B + D
-1
Transfer function =
x s

In general D =0

Lecture Notes by N.Selvaganesan, IIST (January – May 2018)


32

Adj(sI - A)
Transfer function = C B
sI - A

Problem 9.9

Find the transfer function of a SISO system whose A, B, C matrix are given below

1 
 2 -1  m  and C=  -1 1
A=   , B=
 4 -2  1 
 m

Solution

Transfer function =C  [sI-A]-1  B

Computation of [sI-A]-1 :

s-2 1 
sI-A=  
 -4 s+2 
1 s+2 -1 
[sI-A]-1 = 2 
s  4 s+2 

1 
1 s+2 -1   m 
Transfer function =  -1 1 2 
s  4 s-2   1 
 m

 s+2 1 
 m - m 1
=  -1 1   2
 4 + s-2  s
 m m 

 s+1 
=  -1 1  m  * 1 = 1 -  s+1  +  s+2   = 1  -s-1+s+2 
s+2  s 2 s 2   m   m   s 2  m 
 m

1
Transfer function =
ms2

Lecture Notes by N.Selvaganesan, IIST (January – May 2018)


33

Problem 9.10

Find the transfer function of a SISO system whose A, B, C matrix are given below

0 1   0 
A=   , B=  1  and C= 1 0
0 0   m 

Solution

Transfer function =C  [sI-A]-1  B

Computation of [sI-A]-1 :

 s -1
 sI-A  =  
0 s 
1  s 1
 sI-A  = 2  
-1

s 0 s 

1 
1  s 1  0  1
Transfer function = 1 0 0 s   1  s 2 
= 1 0   s m 
s2    m 
 m

1
Transfer function=
ms2

Problem 9.11

Obtain the T.F for a system whose describing equations are given below:


x1 = -sx1 -x 2 +2u

x 2 = 3x1 -x 2 +su
y = x1 +2x 2

Solution

State space representation of the system is given by

Lecture Notes by N.Selvaganesan, IIST (January – May 2018)


34

 
 x1  = -5 -1  x1  +  2  u
    3 -1  x 2   5 
x2 
x 
Y= 1 2  1 
x2 

Transfer function =C  [sI-A]-1  B

Computation of [sI-A]-1 :

s+5 1 
 sI-A  =  
 -3 s+1
s+1 -1  1
sI-A = 
-1
 2
 3 s+5 s +6s+8

1 s+1 -1   2 
Transfer function = 1 2
s +6s+8  3 s+5  5 
2

1  2s+2-5  1  2s-3 
= 1 2   = 1 2 2
2
s +6s+8 6+5s+25 s +6s+8 5s+31
1
=  2s-3+10s+62
(s+4)(s+2)

1
Transfer function= 12s+59
(s+4)(s+2)

Problem 9.12

Determine the T.F of a MIMO system whose A, B and C matrix are given below

-1 0  1 0  1 2
A=   B=   C=  
 0 -2  0 1  4 1 

Solution

Transfer function =C  [sI-A]-1  B

Computation of [sI-A]-1 :

Lecture Notes by N.Selvaganesan, IIST (January – May 2018)


35

s+1 0 
 sI-A  = 
 0 s+2 
1 s+2 0 
sI-A =
-1

(s+1)(s+2)  0 s+1

1 2 1 s+2 0  1 0
Transfer function      
 4 1  (s+1)(s+2)  0 s+1 0 1 

1 2  1 s+2 0 
=   
 4 1  (s+1)(s+2)  0 s+1

 1 2 
1  s+2 2s+2   s+1 s+1 
= 
(s+1)(s+2)  4s+8 s+1   4
Transfer function=
1 
 s+1 s+2 

9.5 Solution of state equations

i) Consider the Homogeneous portion of the equation


x (t)=Ax(t)

By taking Laplace transform

sx(s)-x(0)=Ax(s)

sI-A  x(s)=x(0)  x(s)= sI-A  x(0)


-1

By taking Laplace inverse transform

x(t)=L-1 sI-A  x(0)


-1

x(t) = φ(t) x(0)

where, φ(t) is state transition matrix

I A A2
sI-A = + 2 + 3 +............
-1

s s s

Lecture Notes by N.Selvaganesan, IIST (January – May 2018)


36

A 2 t 2 A3 t 3
φ(t)=L-1[sI-A]-1 =I+At+ + +........= eAt
2! 3!

x(t)=e At x(0)

ii) Consider the non-homogeneous state equation [Forced system]


x (t)=Ax(t)+Bu(t)

Rearranging the equation


x (t)-Ax(t)=Bu(t)

Pre-multiplying on both side by e -At

• d -At
e-At x (t)-e-At Ax(t)=e-At Bu(t)  e x(t)  =e-At Bu(t)
dt 

Integrating both sides with respect to „t‟ between the limits 0 and t

t t
e-At x(t)|0t =  e-Aτ Bu(τ).dτ  e-At x(t)-x(0)=  e-Aτ Bu(τ).dτ
0 0

t
x(t)=eAt x(0)+ e-A(t-τ) Bu(τ).dτ
0

t
x(t) = φ(t)x(0) +  φ(t-τ)Bu(τ).dτ
0

Homogeneous Non-homogeneous
Solution Solution

9.5.1 Properties of state transition matrix

1. φ(0) = eA(0) = 1

2. φ(t) = eAt = (e-At )-1 =[φ(-t)]-1  (or) φ-1 (t)=φ(-t)

Lecture Notes by N.Selvaganesan, IIST (January – May 2018)


37

3. φ(t1 +t 2 ) = eA(t1 +t 2 ) = eAt1 eAt 2 = φ(t1 ) φ(t 2 ) =φ(t 2 )φ(t1 )

4. φ(t 2 -t1 ) φ(t1 -t 0 ) =φ(t 2 -t 0 )

e A(t 2 -t1 ) e A(t1 -t 0 ) =e A(t 2 -t1 +t1 -t0 )


= e A(t 2 -t 0 ) = φ(t 2 -t 0 )

Problem 9.13

A linear time invariant system is characterized by the homogeneous state equation

 . 
 x1   1 0   x1  . Compute the solution of a homogeneous equation. Assume initial
 .  1 1   x2 
 x2 

1 
vector x0   
0 

Solution

Consider the Homogeneous portion of the equation

x =Ax

Solution of the above state equation is x(t)=φ(t).x(0)

Compute φ(t)=L-1 [sI-A]-1

s-1 0 
sI-A=  
 -1 s-1
s-1 0  1
[sI-A]-1 =   2
 1 s-1 (s-1)
 1 
 s-1 0 
 et 0
L-1[sI-A]=L-1  =  t 
 1 1   te et 
 (s-1) 2 s-1 

 et 0  1   e t 
 x(t)=  t  = 
 te e t  0   te t 

Lecture Notes by N.Selvaganesan, IIST (January – May 2018)


38

Problem 9.14

Obtain the State Transition matrix for a second order system describes the equation

 . 
 x1    0 1   x1 
 .   2 3  x2 
 x2 

Solution

Compute φ(t)=L-1 [sI-A]-1

 s -1 
[sI-A]=  
 2 s+3
s+3 1 1
[sI-A]-1 =   2
 -2 s  s +3s+2
 s+3 1 
 (s+1)(s+2) (s+1)(s+2) 
L-1[sI-A]-1 =L-1  
 -2 s 
 (s+1)(s+2) (s+1)(s+2) 

 2e-t -e-2t e t -e-2t 


φ(t) =  -t -2t 
-2e +2e -e -t +2e -2t 

Problem 9.15

Obtain the State vector for the system as described by following differential equation

 
 x1  =  0 1   x1  + 0  and x(0)= 1 
  -2 -3  x 2  1  0
 
x2 

where step function as an Input to the system.

Solution

 2e-t -e-2t e-t -e-2t 


φ(t)=  -t -2t  (Already found in problem 9.14)
-2e +2e -e-t +2e-2t 

Lecture Notes by N.Selvaganesan, IIST (January – May 2018)


39

Method: 1 (Unforced System)

x(s)=[sI-A]-1[x(0)+Bu(s)]

1
u(t)=1  u(s)=
s

1 s+3 1  1  0  1
[sI-A]-1[x(0)+Bu(s)] = × 
(s+1)(s+2)  -2 s   0  1 
+
s
 1
1  s+3+ 
= s
(s+1)(s+2)  
 -2+1 
 s 2 +3s+1 
 s(s+1)(s+2) 
x(t) = L-1  
 -1 
 
 (s+1)(s+2) 

 1 -t 1 -2t 
 s 2 +3s+1  -1  -1  +e - e 
-1
By solving L   and L   , we get x(t)=  2 2
 s(s+1)(s+2)   (s+1)(s+2)   -t -2t 
 -e +e 

Note:

 s 2 +3s+1   -1 
Solution of L-1   and L-1  
 s(s+1)(s+2)   (s+1)(s+2) 

By taking partial fraction and solving the equation

s3 +3s+1=A(s+1)(s+2)+Bs(s+2)+s(s+1)
s=-1  B=-1
s=-2  C=-1/2
s=0  A=1/2

Lecture Notes by N.Selvaganesan, IIST (January – May 2018)


40

-1= A(s+2)+B(s+1)
and s = -1  A=-1 and s = -2  B=1
1 1 1 
 2s + s+1 - 2(s+2) 
x(s)=  
 -1 1 
 s+1 + s+2 

Method: 2 (Forced System)


t
x(t)=φ(t)x(0)+  eA(t-x) Bu(z)dz
0

1
u(s)=  u(t)=1
s

 2e-(t-z) -e-2(t-z) -e-2(t-z)  0 


 e
-(t-z)
t

0 φ(t-z).Bu(z)dz =  -2e-(t-z) +2e-2(t-z) ×


-2(t-z)  1 
  -e +2e   
-(t+z)

t
 e-(t-z) e-2(t-z) 
 e -e dz  
  2 
-(t-z) -2(t-z) -
 = 1
 -e(t-z) +2e-2(t-z) dz   -e-(t-z) 2e-2(t-z) 
   +
 1 2  0

 0 e0  -t e-2t    e-2t -t 1 
e - - e - 
2   =  2
-e + 
= 2  (1)
 -t -2t 2 
 -1+1-[-e +e   e -e
-t -2t]


 2e-t -e-2t 
φ(t)x(0)=  -t -2t  (2)
-2e +e 

 1 -t 1 -2t 
+e - e 
(1)+ (2)  x(t)=  2 2
 -t -2t 
 -e +e 

Lecture Notes by N.Selvaganesan, IIST (January – May 2018)


41

EXERCISE PROBLEMS

1. Formulate the state equation for the given electrical network given in Fig 9.17.

R1 L1 L2

U R2
C

Fig 9.17

2. Formulate the state equation for the given mechanical system shown in Fig 9.18.

B1 K

M1 x1
Fig 9.18
B2

M2 x2

F(t)
3. Draw the state block diagram for the transfer function given below and obtain the state
equations.
C(s) 10(s+3)
a. =
R(s) s(s+1)(s+2)

Lecture Notes by N.Selvaganesan, IIST (January – May 2018)


42

C(s) 2(s+3)
b. =
R(s) (s+1)(s+2)
C(s) 6
c. = 3 2
R(s) (s +9s +26s+24)

 0 1
4. For the A matrix given below fine the state transition matrix A=  
-3 4 
SOLUTION

 R1 1 1 
   L 0 -  L 0
L1
 I1   1   I1   2 
   1  1   V1 
-   I 2  +  0
R2
1. State equation is  I 2  =  0
L 2   V2 
-
   L2 L2
  vc   
 vc   1 1  0 0
   0   
C C   

Output equation is given by y = I1

 I1 
y = 1 0 0 I 2 
 vc 

   0 0 1 0 
 x1   0  x   0 
  
0 0 1  1  0 
 k  x2   
2. State equation is    = -
x2 B B2
0 - 2   +  0   F
  M M1 M1 x  
x3   1   3
 1 
   0 B2 -(B1 +B2 )   x 4 
 M 2 
 x 4   0 
 M2 M2 

Output equation is given by y = x 2

 x1 
x 
y = 1 0 0 0  2 
x3 
 
x4 

Lecture Notes by N.Selvaganesan, IIST (January – May 2018)


43

 40 
0 0 0  3 
 
3. a) A = 0 -1 
0  , B   15 and C  1 0 0

0 0 3  5 
 
 3 

 1 0  4
b) A =   , B    and C  1 1
 0 2   2 

-2 0 0  0.75 
c) A =  0 -3 0  , B   1.5 and C  1 1 1


 0 0 4  0.75 

 1.5e-t -0.5e-t 0.5e-t -0.5e -3t 


4. x(t)=  
 1.5e  1.5e
-t -3t
-0.5e-t +1.5e -3t 

Lecture Notes by N.Selvaganesan, IIST (January – May 2018)

Das könnte Ihnen auch gefallen