Professional Documents
Culture Documents
of Control Systems
Output Equation: y t Cx t Du t
x= state vector,
u = input vector,
y = output vector and
A, B, C, D are the respective matrices.
Relationship Between Transfer Function and State
Space Equation:
Y ( s)
G s
U (s)
State Model: x Ax Bu y Cx Du
⇒ sX( s) x(0) AX( s) BU( s)
Introduction
Y( s) CX( s) DU( s)
With zero initial condition {for transfer function}:
1
(sI A)X(s) BU(s) ⇒ X(s ) (sI A) BU(s)
System Equation:
m y b y ky u
Introduction
• This is a 2nd order system, which means that there
are two integrators.
Let x1 (t ) y(t ) and x2 (t ) y (t )
Thus, x1 x2
1 1 k b 1
x2 ky by u x1 x2 u
m m m m m
State Eq: 0 1 0
x1 k x1
b 1 u
x x2
2 m m m
x1
Output Eq: y 1 0
x2
Introduction
,
Comparison with standard form of State Space Model:
x1 0 1 0
x A k b B1
C 1 0 D0
x2
m m m
Block Diagram Model:
Introduction
G(s) C(sI A)1 B D
1
0 1 0
s 0
1 0 k
b 1 0
0 s
m m m
1
2
ms bs k
State Space Model of n-th order system
(Input doesn’t have derivative terms)
System Dynamics:
( n 1)
y (n)
a1 y an1 y an y u
( n 1)
State Variables: x1 y, x2 y, ..., xn y
So, x1 x2 x2 x3
xn1 xn
xn an x1 an1 x2 a1 xn u
State Eq: x Ax Bu
x1 0 1 0 0 0
x2 0 0 1 0 0
x A B
0 0 0 1 0
xn an an1 an2 a1 1
x1
x2
Output Eq: y 1 0 0 ⇒ y Cx Du
xn
C 1 0 0 D 0
Y ( s) 1
Also n
U (s) s a1s n1 an1s an
State Space Model of n-th order system
(Input function has derivative terms)
System Dynamics:
( n1) ( n1)
y a1 y
( n)
an1 y an y b0u b1u ( n)
bn1 u bnu
• State variables must be chosen so that they
eliminate the derivatives of u in the state equation.
0 b0 1 b1 a10 2 b2 a11 a2 0
n bn a1n1 an11 an 0
Thus, x1 x2 1u x2 x3 2u
x n1 xn n1u x n an x1 an1 x2 a1 xn nu
State Eq: x Ax Bu
1 x1
0 1 0 0 2 x2
0 0
A
0 1
B x
0 0 0 1
n1
an an1 an2 a1 n xn
x1
x2
Output Eq: y 1 0 0 0u
xn
x1 p1 0 0 0 x1 1
x 2 0 p2 0 0 x2 1
0 0 0
u
x 0 0 0 pn xn 1
n
x1
x2
y c1 c2 cn b0u
xn
(4) Jordan Canonical Form:
• Denominator polynomial of transfer function
involves multiple order poles.
Example: Y ( s) b0 s n b1s n1 bn1s bn
U ( s ) ( s p1 )3 ( s p4 ) ( s pn )
c1 c2 c3 c4 cn
b0
( s p1 ) ( s p1 ) ( s p1 ) ( s p4 )
3 2
( s pn )
x1 p 1 0 0 0 x1 0
x2 1
x2 0
0 p1 1
x3 0 0 p1 0 0 x3 1
u
0 0 p4 0 0 x4 1
x4 0 0
0 0 0 0 pn xn 1
x n
x1
x2
y c1 c2 cn b0u
xn
Example:
Y ( s) s3
2
U ( s ) s 3s 2
(1) Controllable Canonical Form:
x1 0 1 x1 0 u
x 2 2 3 x2 1
x1
y 3 1
x2
(2) Observable Canonical Form:
x1 0 2 x1 3 u
x 2 1 3 x2 1
x1
y 0 1
x2
(3) Diagonal Canonical Form:
x1 1 0 x1 1 u
x 2 0 2 x2 1
x1
y 2 1
x2
Eigenvalues of Matrix A :
Eigenvalues of A: Roots of the characteristic equation.
λI A 0
Diagonalization of Matrix A:
x Ax Bu y Cx
Matrix A, having distinct eigenvalues, is:
0 1 0 0
0 0 1 0
A
0 0 0 1
an an 1 an 2 a1
Transformation: x Pz
1 1 1
1 2 n
2 2 {where λ1,λ2,…,λn are the
P 1 22 n
distinct eigenvalues of A}
1n1 2n1 nn1
1 0 0 0
1
0 2 0 0
P AP
⇒ 0 0 0
0 0 0 n
Pz APz Bu
z P1APz P1Bu
y CPz
Example:
0 1 0
A 0 0 1 has eigenvalues λ1, λ1, and λ3.
a a2 a1
3
1 0 1
x Sz S 1 1 3
2 2
1 21 3
1 1 0
S 1AS 0 1 0
0 0
3
Invariance of Eigenvalues:
• Invariance of the eigenvalues under a linear
transformation.
1
• Characteristic polynomial λI A and λI P AP
are identical.
Proof: 1 1 1
λI P AP λP P P AP
P 1 (λI A)P
P 1 λI A P
P 1 P λI A P 1P λI A
λI A
Solution of Homogeneous State Eqs:
Scalar differential Eq: x ax
Let us assume a solution,
x t b0 b1t b2t 2 bkt k
On substitution,
b1 2b2t 3b3t 2 kbk t k 1
a(b0 b1t b2t 2 bkt k )
Thus, 1 1 2 1 1 3
b1 ab0 , b2 ab1 a b0 b3 ab2 a b0
2 2 3 3 2
1 k
bk a b0 and b0 x(0)
k!
1 22 1 33
So, x(t ) 1 at a t a t 1 k k
at x(0)
2! 3! k!
e at x 0
1 22 1 k k
⇒ x t I At A t At x 0
2! k!
e x 0
At
Matrix Exponential:
k k
A t
e
At
k 0 k !
d At 1 22 Can be easily
e I At A t A e At
A
dt 2! verified.
---------------------------------------------------------------------
A (t s )
e e e
At As
Proof:
A k t k A k s k k t i s k i
e At e As A i ! ( k i )!
k 0 k ! k 0 k ! k 0 i 0
( t s ) k
Ak eA( t s )
k 0 k!
• If s t , e At At
e e A( t t )
I
At
So, inverse of e e
At
---------------------------------------------------------------------
( A B) t
e e e if At Bt
AB BA
and e( A B)t e At eBt if AB BA
Proof:
A B
2
e ( A B) t
I A B t t2
2!
A 2 2
t B 2 2
t
e e I At
At Bt
I Bt
2! 2!
A 2t 2 B 2t 2
I A Bt ABt
2
2! 2!
Thus,
( A B) t BA AB 2 BA 2 ABA B 2 A BAB 2A 2B 2AB 2 3
e e e
At Bt
t t
2! 3!
Verification:
x 0 Φ 0 x 0 x 0
and x(t ) Φ t x 0 AΦ t x 0 Ax t
Φ t e At L1[( sI A) 1 ]
⇒ Φ 1 t e At Φ t
Φ t ≡ State transition matrix
1t 2t nt
e ,e , ,e
1t
Specifically, e 0 0 0
0 e2t 0 , if A is diagonal.
Φ t e
0
At
0 0 0
0 0 0 ent
Φ t1 t2 e eAt1 eAt2 Φ t1 Φ t2 Φ t2 Φ t1
A t1 t2
(3)
(4) Φ t Φ nt
n
2et e2t et e 2 t
Φ t Φ t
1
2e t
2e 2t
et 2e2t
Solution of Nonhomogeneous State Equations:
Scalar Case: x ax bu ⇒ x ax bu
• d at
⇒ e at
x t a x t
dt [e x t ] e at
bu (t )
t
⇒ e at x t x 0 e a bu ( )d
0 t
⇒ x t x 0 eat e at e a bu ( )d
0
X s sI A x 0 sI A B U( s )
1 1
⇒
L e At x 0 L e At B U( s )
t
⇒ x t e x 0 e A (t )Bu( )d
At
λI AλI A
Proof: 1
I
Using Cramer’s rule for inverse of any matrix (say Y):
y11 y12 y1n
1 1
Y
Y y yn 2 ynn
n1
where y jk
p11 ( ) p1n ( )
So, 1
λI A
1
p A (λ) p ( ) pnn ( )
n1
where all
Thus, we can write,
pA λ λI A λn1Bn1 λn2Bn2
1
λB1 B0
n n matrices.
⇒ p A λ I λI A λ Bn1 λ Bn2 λB1 B0
n 1 n2
A
n 1
Let p λ λ n
c n 1λ c1λ c0
This is true if matrices used with each power of λ is an
identity matrix multiplied by respective coefficients.
Then, by comparison,
Bn1 I Bn2 ABn1 cn1I ,…, B0 AB1 c1I and AB0 c0I
p A A A n cn1A n1 cn2 A n2 c1A c0 I
A nB n1 A n1 B n2 AB n1 A (B 0 AB1 ) AB 0
0
c1 c2 cn 0
n
xi c j x j
j 1
j i
Controllability and Observability
• A system is said to be controllable at time t0, if it is
possible by means of an unconstrained control
vector to transfer the system from any initial state
x(t0) to any other state in a finite time interval.
• A system is said to be observable at time t0, if with
the system state at x(t0), it is possible to determine
the state from the observation of the output over a
finite time interval.
Then, x 0 e Bu d
A t
x t e At
0
t
y t Ce At x 0 Ce Bu d Du
A t
0
• A, B, C, and D are known. u(t) is also known. Hence
the last two terms of y(t) equation are known
quantities. Thus, these two terms can be subtracted
from the observed values of y(t).
• Hence, for investigations on necessary and
sufficient conditions for complete observability, it is
enough to consider the unforced system.
⇒ It can be proved that x(0) can be uniquely
determined if and only if the rank of
C
CA
is n.
CA
n1
C* A*C* (A* )n1C* = Observability Matrix
nnm