Sunteți pe pagina 1din 89

State Space Analysis of C Control l system

Topics

Ap practical control system. y State space representation Definitions. Critical considerations while selecting state variables. State variable selection. Advantages of state variable representation. Generic state space representation. Block diagram representation of linear systems. Writing differential equations in First Companion form
State Space Analysis of Control System Mugdha Salvi, VCET 3

29-Apr-13

A practical control system


(bi noise) (bias i )

Window Wi d opening/closing i / l i (random noise)

29-Apr-13

State Space Analysis of Control System Temperature control system in a car Mugdha Salvi, VCET

Another Practical Control System

noise

Water level control in an overhead tank


29-Apr-13 State Space Analysis of Control System Mugdha Salvi, VCET 5

State Space Representation


noise

Input variable:
Manipulative (control) Non-manipulative p ( (noise) )

control

Y Y* Z

System Controller

Output variable:
Variables of interest that can be either be measured or calculated

State variable: Minimum set of parameters which completely y status. summarize the systems
State Space Analysis of Control System Mugdha Salvi, VCET 6

29-Apr-13

Definitions
State: The state of a dynamic system is the smallest number of variables (called state variables) such that the knowledge d of these variables b at t = t0, together with the knowledge of the input for t t0, completely determine the behavior of the system for any time t t0. Note: State variables need not be physically measurable or observable quantities. This gives extra flexibility.
29-Apr-13 State Space Analysis of Control System Mugdha Salvi, VCET 7

Definitions
State vector: A n - dimensional vector whose components are n state variables that describe the system completely. St t Space: State S the n - dimensional space whose co-ordinate axes consist of the x1 axis, , x2 axis, , ., , xn axis is called a state space. Note: N t For any dynamical system, the state space remains unique, q , but the state variables are not unique. q
29-Apr-13 State Space Analysis of Control System Mugdha Salvi, VCET 8

Critical Considerations while selecting State Variables.


Minimum number of variables


Minimum number of first-order differential equations needed to describe the system dynamics completely Lesser number of variables: won wont t be possible to describe the system dynamics Larger number of variables:

Co p Computational o co complexity pe y Loss of either controllability, or observability or both.

Linear independence. If not, it may result in:


Bad: May not be possible to solve for all other system variables Worst: May not be possible to write the complete state equations
State Space Analysis of Control System Mugdha Salvi, VCET 9

29-Apr-13

Sate Variable Selection


Typically, the number of state variables (i.e. the order of the system) is equal to the number of p energy gy storage g elements. However, independent there are exceptions! Is there a restriction on the selection of the state variables ? YES! All state variables should be linearly independent y must collectively y describe the system y completely. p y and they
State Space Analysis of Control System Mugdha Salvi, VCET 10

29-Apr-13

Advantages of State Space R Representation i


Systematic y analysis y and synthesis y of higher g order systems without truncation of system dynamics Convenient tool for MIMO systems U if Uniform platform l tf for f representing ti time-invariant ti i i t systems, time-varying systems, linear systems as well as nonlinear systems Can describe the dynamics in almost all systems (mechanical systems, electrical systems, biological systems, y , economic systems, y , social systems y etc.) ) Note: Transfer function representations are valid for only l f for linear li time i invariant i i (LTI) systems
State Space Analysis of Control System Mugdha Salvi, VCET 11

29-Apr-13

State variable representation of a system

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

12

Generic State Space Representation

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

13

Generic State Space Representation

Summary:

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

14

Block diagram representation of linear systems


& x (t) = Ax(t ) + Bu (t ) : state equation y (t ) = Cx (t ) + Du (t ) : output equation

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

16

State Model for Linear system


u Input
SISO system Initial state : x(0)

y Output

State variables

& x (t) ( ) = Ax(t ) + bu (t ) : state equa q tion y (t ) = cx (t ) + du (t ) : output equation x1 (t ) a11 a12 a x (t ) a22 2 21 ;A = x(t) = x ( t ) n an1 an 2 y (t ) = x(t )
29-Apr-13

a1n b1 b a2 n ; b = 2 ; c = [c c 1 2 ann bn

cn ]

State Space Analysis of Control System Mugdha Salvi, VCET

17

State Model for Linear system


& x (t) = Ax(t ) + Bu (t ) : state equation u1 - up MIMO system y (t ) = Cx (t ) + Du (t ) : output p equ q ation Input Initial state : x(0) x1 (t ) a11 a12 a1n a x (t ) a a 22 2n x( (t) ) = 2 ; A = 21 ; State variables x ( t ) a a a n2 nn n n1 b11 b12 b1 p c11 c12 c1n d11 d12 b c d b b c c d 22 21 22 2p 21 22 2n 21 B= ;C = ;D = b b b c c c n1 n 2 q1 q 2 np qn d q1 d q 2 y (t ) = x(t )
y1 - yq Output

d1 p d2 p d qp

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

18

Writing Differential Equations in First Companion Form (Phase variable form/Controllable canonical form)

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

19

First Companion Form (Controllable Canonical Form)

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

20

Example 1
First Companion Form (Controllable Canonical Form)

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

21

Example 2 (spring(spring -mass mass-damper system)

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

22

Example 3 (R L C circuit)
First Companion Form (Controllable Canonical Form)

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

23

Realization of First Companion Form (Controllable Canonical Form)

Consider only the following transfer function

Corresponding differential equation is: where, Solving for highest derivative of z(t) we obtain

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

24

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

25

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

26

First Companion Form (Controllable Canonical Form)

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

27

Example 4 (Constant as numerator)

By cross multiplication:

The corresponding differential equation is found by taking the inverse Laplace transform, assuming zero initial conditions:

Choose state variables as:


State Space Analysis of Control System Mugdha Salvi, VCET

29-Apr-13

28

Example 4

Vector matrix form it can be written as:

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

29

Example 4

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

30

Example 5 (T.F. T with polynomial in numerator)

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

31

Example 5
Introduce the effect of the block with the numerator., where b1 =1; b2= 7, and b3=2, states that

C ( s ) = ( 1 s 2 + 2 s + 3 ) X 1 ( s )
Taking Inverse Laplace Transform:

But

Hence, Hence

y (t ) = c(t ) = 1 x3 + 2 x2 + 3 x1 = x3 + 7 x2 + 2 x1

[1
29-Apr-13

3 ]

State Space Analysis of Control System Mugdha Salvi, VCET

32

Example 5

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

33

Second Companion Form / Observable Canonical Form

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

34

Second Companion Form (Observable Canonical Form)

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

35

Second Companion Form (Observable Canonical Form)

We observe that A, b and c matrices of one companion form corresponds to the transpose of the A, c and b matrices, respectively of the other.
State Space Analysis of Control System Mugdha Salvi, VCET

29-Apr-13

36

Diagonal Canonical Form


The poles of the transfer function appear in the main diagonal This form follows directly from the partial fraction expansion of the transfer function

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

37

Case 1: Poles are Real and Distinct

The matrix is a diagonal matrix with the poles of G(s) as its diagonal elements l
29-Apr-13 State Space Analysis of Control System Mugdha Salvi, VCET 38

Case 2: Real and Complex poles

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

40

Case 3: Real and Repeated poles


Partial fraction expansion is given by:

where

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

41

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

42

Case 3: Real and Repeated poles

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

43

Jordan Canonical Form

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

45

Jordan Canonical Form

Where each of the submatrices i is in the Jordan form . The b and c matrices of the overall system are the concatenations of the bi and ci matrices respectively of each of the subsystems:

The state variable model derived for the case of distinct poles, is a special case of Jordan canonical form where each Jordan block is of 1 x 1 dimension
29-Apr-13 State Space Analysis of Control System Mugdha Salvi, VCET 46

Jordan Canonical Form

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

47

Example 6
Second Companion Form (Observable Canonical Form)
Obtain state-space representations in the controllable canonical form, observable canonical form, and diagonal canonical form form.

controllable canonical form

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

49

Example 6
observable canonical form

diagonal canonical form

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

50

Converting from State Space to a Transfer Function Transfer


Given the state and output equations take the Laplace transform assuming zero initial conditions:

Solving for X(s) or Substituting X(s)into Y(s)yields

where I is the identity matrix.

if U(s) = U(s) and Y(s) = Y(s) are scalars, we can find the transfer function,

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

51

Example - 7
Given the system defined below, find the transfer function, T(s) / Y(s)=U(s), where U(s) is the input and Y(s) is the output.

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

52

Example - 7
Calculate (sI - A)-1:

Therefore, the transfer function is:

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

53

Eigenvalues

The roots of the characteristic equation are referred to as eigenvalues of the matrix A.
(i I A) = 0 ( sI A) = 0

which is called the characteristic equation of the system.

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

54

Properties of eigenvalues
1. 2.

If coefficients of A are all real, , its eigenvalues g either real or in complex conjugate pairs. If 1, 2, , n are the eigenvalues of A, then
tr(A) = i
i =1 n

i e the trace of A is sum of all the eigenvalues of A. i.e. 3. If i, i = 1, 2, , n is an eigenvalue of A, it is an eigenvalue of A'. 4. If A is nonsingular, with eigenvalues of i, i = 1, 2, , n , then 1/ i, i = 1, 2, , n , are the eigenvalues of A-1.
29-Apr-13 State Space Analysis of Control System Mugdha Salvi, VCET 55

Eigenvectors

Any nonzero vector xi that satisfies the matrix equation


Ax i = i x i (i I A) x i = 0

where i, i = 1, 2, , n , denotes the ith eigenvalue of A, is called the eigenvector of A associated with the eigenvalue i. If A has distinct real roots, then A can be diagonalized by the transformation matrix P whose basis vectors are the eigenvectors g of A.
State Space Analysis of Control System Mugdha Salvi, VCET 56

29-Apr-13

Generalized Eigenvectors

If A has multiple order eigenvalues and is nonsymmetric. nonsymmetric . Assume that A has q (<n) distinct eigenvalues, eigenvalues then q eigenvectors are given by
(i I A) x i = 0, where h i is i ith eigenvalue i l , i = 0,1,..., 01 q

Among remaining highhigh-order eigenvalues, let j be of the mth order (m n q); corresponding generalized eigenvectors are given by:
State Space Analysis of Control System Mugdha Salvi, VCET 57

29-Apr-13

Generalized Eigenvectors
( j I A) x n-q +1 = 0 ( j I A) x n-q + 2 = x n-q +1 ( j I A) x n-q +3 = x n-q + 2 ( j I A) x n-q + m = x n-q + m 1

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

58

Example 8 (Eigenvectors)
Find the eigenvectors of the matrix

the eigenvalues are = -2, and -4

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

59

Example 8

Thus, Th

Using the other eigenvalue, -4, we have Thus, eigenvectors is

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

60

Diagonalization I
Note that if an n x n matrix A with distinct eigenvalues is given by:

1, 2, , n = n distinct eigenvalues of A the transformation x = Pz, where

will transform P-1 AP into the diagonal matrix, or

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

61

Example 9 (Diagonalization I)
Consider the following state-space representation of a system:

Eigen values are:

If we define a set of new state variables zl, z2, and z3 by the transformation

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

62

Example 9 (Diagonalization I)

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

63

Example 9 (Diagonalization I)

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

64

Transformation to Jordon Canonical Form


If the matrix A involves multiple eigenvalues, then diagonalization is impossible. But it is then transformed into the Jordan Canonical Form.

has the eigenvalues 1, 1, 3 then the transformation x = Sz, where

will yield

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

65

Transformation of State variables


The original g system y dynamics y are given g by, y


& x (t) = Ax(t ) + bu (t ) ; x(t0 ) = x 0

Th on transformation Then f i the h system d dynamics i are d described ib d as: -1


x (t) = A x (t ) + bu (t ) ; x (t0 )P x (t0 ) y (t ) = c x (t ) + du (t ) where, A = P -1AP, AP b = P -1b, b c = cP, cP d = d

y (t ) = cx (t ) + du (t )

(Note: Both systems have identical output response to the same input) p )
State Space Analysis of Control System Mugdha Salvi, VCET 66

29-Apr-13

Diagonalization II
If A has distinct real roots, then A can be diagonalized by the transformation matrix P whose basis vectors are the eigenvectors of A.

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

67

Example 10 (Diagonalization II)


System is given by:

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

68

Example 10 (Diagonalization II)

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

69

Solution of Homogeneous state equation in Time T domain


We can write the solution of the homogeneous state equation as or where (t) is an n x n matrix and is the unique solution l i of f Also,

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

70

Solution of Homogeneous state equation in Laplace domain


We can write the solution of the homogeneous state equation as or The matrix exponential is computed as,

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

71

Solution of Non Homogeneous state equation in Time domain


Consider d the h non homogeneous h state equation: The solution in the time domain is given by:

Where by definition, and which is called the statetransition matrix.


29-Apr-13 State Space Analysis of Control System Mugdha Salvi, VCET 72

The first term on the right rightg -hand side of the equation q is the response due to the initial state vector, x(0). Notice also that it is the only term dependent on the initial state vector and not the input. We call this part of the response the zero zero-input response, since it is the total response if the input is zero. Th second The d term term, , called ll d the h convolution l i i integral l, is i dependent only on the input, u, and the input matrix, matrix, B, not the initial state vector. We call this part of the response the zero zero-state response, since it is the total response if the initial state vector is zero.
State Space Analysis of Control System Mugdha Salvi, VCET 73

29-Apr-13

Consider the state and output equation:

Solution of Non Homogeneous state equation in Laplace domain

Taking the Laplace transform of both sides of the state equation yields OR

Taking the Laplace transform of the output equation yields


29-Apr-13 State Space Analysis of Control System Mugdha Salvi, VCET 74

Taking g the Laplace p Inverse of the state equation q yields y

Where, Where

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

75

Properties of State Transition Matrix (STM) T


For the time-invariant system: For which We have the following:

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

76

if the matrix A is diagonal, then

the h eigenvalues, i l 1, 2, , n of f the h matrix i A are distinct, di i

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

77

If there is a multiplicity in the eigenvalues, eigenvalues for example, if the eigenvalues of A are: 1, 1, 1, 4, 4 5, 5 , n then (t) will contain, contain in addition to the exponentials e1t, e4t, e5t, , ent , terms like te1t, t2e1t .

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

78

Cayley Hamilton Theorem


The Cayley-Hamilton theorem is very useful in proving theorems involving matrix equations or solving problems involving matrix equations. C id an n x n matrix Consider i A and d its i characteristic h i i equation: i The Cayley-Hamilton theorem states that the matrix A satisfies its own characteristic equation, or that

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

79

Proof for the Cayley Hamilton Theorem Theorem


To prove this theorem, note that adj( I - A) is a polynomial in A of degree n - 1. That is, where B1 = I. Since we obtain

From this equation, we see that A and Bi (i = 1,2,. . . , n) commute. Hence, the product of ( I - A) and adj( I - A) becomes zero if either of these is zero. If A is substituted for in this last equation, then clearly I - A becomes zero. Hence, we obtain
This proves the Thi th Cayley-Hamilton C l H ilt theorem
29-Apr-13 State Space Analysis of Control System Mugdha Salvi, VCET 80

Methods for computing

At e

Method 1: If matrix A can be transformed into a diagonal form, then eAt can be given by:

where P is a diagonalizing matrix for A. If matrix A can be transformed into a Jordan canonical form, then eAt can be given by

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

81

Methods for computing


Method 2:

At e

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

82

Methods for computing

At e

Method 3: The third method is based on Sylvester's interpolation method. We shall first consider the case where the roots of the minimal polynomial ( ) of A are distinct.. Solve the determinant to find value of eAt about the last column

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

83

Sol ing the determinant is similar to solving Solving sol ing the equation eq ation below: belo

Where, i can be found by solving the following equations simultaneously,

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

84

the case where the roots of the minimal polynomial ( ) of A are multiple.. Solve the determinant to find value of eAt about the last column

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

85

the case where the roots of the minimal polynomial ( ) of A are multiple..

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

86

Example 11 (STM and x(t))


Consider the following matrix A: Compute eAt the 3 methods
Method 1:

The eigenvalues of A are 0 and -2 (1, = 0, 2 = -2). A necessary transformation matrix P may be obtained as

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

87

Example 11
Method 2:

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

88

Example 11
Method 3:

Substituting b 0f for 1, and d -2 for 2 in this equation, we obtain Expanding the determinant, we obtain

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

89

Example 11
Method 3:

Since 1, = 0 and 2 = -2, , the above two equations q become

Solving for 0(t) and 1 (t) gives

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

90

For the state equation and initial state vector shown find the state-transition matrix and then solve for x(t), x(t) where u(t) is a unit step
and

Example 12 (STM) T

To find STM which is given by, Calculate ( (sI A) )-1

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

91

Example 12
taking g the partial p fractions

taking the inverse Laplace transform of each term, we obtain

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

92

Example 12

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

93

Example 12

29-Apr-13

State Space Analysis of Control System Mugdha Salvi, VCET

94

S-ar putea să vă placă și