215461: Automatic Flight Control I

Lecture for Week12

State Space Representations of Dynamic
Systems: continued

Sept. 9, 2013 Aerospace Engineering, 1

 d
x(t ) Ax(t ) Bu (t ) State equation
Dynamic equation dt
y (t ) Cx (t ) Du (t ) Output equation
State variable
x1 (t ) u1 (t ) y1 (t ) x1 (0)
x (t ) u (t ) y (t ) x (0)
x(t ) 2 u (t ) 2 y (t ) x(0) 2
2

x n (t ) n1 ur (t ) r1 y (t )
p p1 x ( 0)
n n1

State space r- input p- output

A nn
B n r
C pn
D pr

Sept. 9, 2013 Aerospace Engineering, 2

 u1 (t ) y1 (t )
u2 (t ) Inner state variables y2 (t )
x1 , x 2 , x n
u r (t ) y p (t )

u (t ) + x (t ) 1 x(t ) + y (t )
B C
s +
-

Sept. 9, 2013 Aerospace Engineering, 3

 State Space Models
state-space description

transfer ftn state-space eq.

pole/zero ? canonical form
eigenvalue/eigenvector
important concept
controllability pole-location controller
observability estimator(observer)
(cf) pole/zero concellation
stability(Lyapunov sense)

internal model principle

Sept. 9, 2013 Aerospace Engineering, 4

 Advantages of State Space
can study the internal structure of systems

x(t ) Ax(t ) Bu (t ) y ( s ) G ( s )u ( s )

y (t ) Cx (t ) Du (t )
consider only I/O characteristics
input state output
space B space C space
A

What is the state? D

System dynamics describe minimum number of
vectors
Lets consider a well-know system, e.g. electrical circuit.

Sept. 9, 2013 Aerospace Engineering, 5

 Example: RLC electrical circuit
t=0 R L

+
physical law
vi(t) i(t) Vc(t) i L (o ) 0, vc (o ) 0
d 1 t
-
vi (t ) Ri (t ) L i (t ) i (t )dt
dt c 0

Now we will derive the differential equation

t
define (t ) i (t )dt
0 0 1 x 0
x1 1
R 1 v
x2 i
1
1


vi (t ) R L 2nd order
c x2 LC L L
consider vc (t ) as output
define x1 , x2 1 1 1
y vc i dt x1
c c c
the x1 x2

1 R 1 x(t ) Ax(t ) Bu (t )
x1 vi y (t ) Cx (t )
LC L L
0 1 0
R , B 1 , C
1 R 1 1
x1 x2 vi where A 1 0
C
LC L L LC L L

Sept. 9, 2013 Aerospace Engineering, 6

Motivation of state space approach

Example 1 e(t )
u (t ) s 1 y (t )
+ - ( s 1)

2
s 1 +
n(t ) noise
Y (s) s 1
Transfer function
U ( s ) s 3 BIBO stable

E ( s) 2( s 1)
u 0
N ( s) ( s 1)( s 3)
unstable

Sept. 9, 2013 Aerospace Engineering, 7

Example 2

u (t ) s 1 1 y (t )
s 1 s 1

y 1
BIBO stable, pole-zero cancellation
u s 1

+
+ x 1 1 x1 + v(t ) x 2 1 x2
-2
u (t ) - s + s y (t )
+
v s 1 2
1
u s 1 s 1
2
vu u
s 1
Sept. 9, 2013 Aerospace Engineering, 8
 x 1 x1 2u x1 (0) x10
x 2 x1 x2 u x 2 (0) x 20
y x2
x(t ) L1 [( sI A) 1 ]x(0) L1 [( sI A) 1 BU ( s)]

x1 (t ) e t x10 2e t u (t )
1 1 t
y (t ) x 2 (t ) ( x 20 x 10 )e e x10 e t u (t )
t

2 2

1. if x10 x 20 0
2. if x10 2 x 20 then system stable

State-space description Internal behavior description

Sept. 9, 2013 Aerospace Engineering, 9

 Definition: The state of a system at time t 0 is the amount of
information at t 0 that together with u[t 0 , ] determines
uniquely the behavior of the system for t t 0

Example x(t )

M f (t )

f (t ) Ma Mx (t )

With all nessary

information:
f [t0 , ] t 0 x(t 0 ) x (t 0 )

x(t ) x (t ) can be determined t t0

Sept. 9, 2013 Aerospace Engineering, 10

Example : Capacitor electric
energy
1 t0 1 t
v(t ) i ( ) d i ( )d
c c t0
Input i[, t 0 ]
1 t
v (t 0 ) i ( )d

c t0

Example : Inductor

1 t0 1 t
i (t ) v( )d v( )d
L L t0
1 t
i (t 0 ) v( )d

L t0 Magnetic
energy
Sept. 9, 2013 Aerospace Engineering, 11
Example y2 y1
K
M2 M1 f (t )

B3 B2 B1

M 1 y1 B1 y 1 B2 ( y 1 y 2 ) K ( y1 y 2 ) f (t )
M 2 y 2 B3 y 2 B2 ( y 2 y 1 ) K ( y 2 y1 ) 0

x1 y1 y 2 x1 0
1 1 x1 0
x K ( B B2 ) B2 x 1 f (t )
let x 2 y 1 2 M 1
M1 M1 2 M1
x 3
( B2 B3 ) x3 0
1
x3 y 2
K B2

M 2 M2 M 2

Sept. 9, 2013 Aerospace Engineering, 12

Example La Rf
Ra

ea t ia e t Lf ef t
b

Armature circuit m Field circuit
TL
Jm
di Tm k i ia
ea (t ) Raia eb La a
dt d 2 m d m
Jm B TL
eb kb m dt 2 m
dt

dia Ra kb 1 d m k B d m 1
ia m ea i ia m TL
dt La La La dt Jm J m dt Jm
d m
m
dt

Sept. 9, 2013 Aerospace Engineering, 13

 Ra Kb 1
L La
0 L 0
ia a ia a e
Ki Bm 1 a
m J Jm
0 m 0

J m TL
m m
0 1 0 m 0 0


ia
y1 (t ) m (t ) 0 0 1
y (t ) (t ) 0 1 0 m
2 m
m

Sept. 9, 2013 Aerospace Engineering, 14

 Dynamical equation Transfer function

x (t ) Ax(t ) Bu (t )
y (t ) Cx(t ) Du (t )
Laplace transform
sX ( s ) x(0) AX ( s ) BU ( s )
Y ( s ) CX ( s ) DU ( s )

assume x(0) 0
X ( s ) ( sI A) 1 BU ( s )
Y ( s ) [C ( sI A) 1 B D]U ( s )
matrix
Transfer function

Sept. 9, 2013 Aerospace Engineering, 15

 Example By Newtons Law
F My My by ky r
r ky by My let x1 y, x2 y
k

b
x 1 y x 2
M
b k 1
x 2 y y y r
y (t ), y (t ) M M M
r (t ) b k 1
x2 x1 u u r
M M M
x 1 x 2
k b 1
x 2 x1 x2 u
M M M

x 1 0 1 x 0
x
k b 1 1 u y 1 0 1 0 u
x 2 M x2
M M x2
C
A B

Sept. 9, 2013 Aerospace Engineering, 16

 State Space Equation
x 1 0 1 x 0
For example : k b 1 1 u
x Ax Bu
x 2 M x2
M M
y Cx Du
x1
y 1 0 0 u
x2
Transfer Function
Y (s) b1s b0
G(s) For example : G ( s )
U (s) a2 s 2 a1s a0

Example: Transfer function of the Mass-damper-spring system

d2y dy
M 2 b ky u (t ) x x1 x2
T
dt dt
Ms 2Y ( s ) bsY ( s ) kY ( s ) U ( s )
Y (s) 1
G(s)
U (s) Ms 2 bs k

Sept. 9, 2013 Aerospace Engineering, 17

 x 1 0 1 0 x1 0 0
Example x 0 4 3 x 1 0 u1
2 2 u
x 3 1 1 2 x3 0 1 2
MIMO system x1
y1 (t ) 1 0 0
y (t ) 0 0 1 x2
2 x
3

adj ( sI A)
( sI A) 1
sI A
s 2 6 s 11 s 2 3
1
3 s 2
2 3s
s( s 4)( s 2) 3 3s
s4 s 1 s 4 s
2

G ( s) [C ( sI A) 1 B D ]
1 s2 3
Transfer function

s 3 6 s 2 11s 3 ( s 1) s ( s 4)

Sept. 9, 2013 Aerospace Engineering, 18

 How many state variables?
Natural Question
Is the state-space description unique?
(Answer) No.

We can select in the above example

1
x1 i dt , x 2 1
C
Note
1. State-space decryption is not unique.
(the state is not unique)
2. The state can have a physical meaning or not.
This means we can select the state in the way the that make the
controller design easier when the physical interpretation of the state is
not important.

Sept. 9, 2013 Aerospace Engineering, 19

 Remark : the choice of states is not unique.

+ R L +
di (t ) 1 t
ei (t ) c ec (t ) Ri (t ) L i (t )dt ei (t )
i (t )
dt c 0
- -

x1 (t ) i (t ) x 1 R
1 x 1
x L LC L ei (t )
1
let 2 1 0 x2 0
x2 (t ) i (t )dt
x
y (t ) i (t ) y (t ) 1 0 1
x2
R R
x1 (t ) i (t ) x1 L

L
x 1
L ei (t )
1
let 1
x 2 (t ) ec (t ) x 2 0 x2 0
exist a mapping L
y (t ) i (t ) x1
y (t ) 1 0
x 2
Sept. 9, 2013 Aerospace Engineering, 20
 Different state equation description
x(t ) Ax(t ) Bu (t ) x (t ) A x (t ) B u (t )
y (t ) Cx(t ) Du (t ) x(t ) px (t ) y (t ) C x (t ) D u (t )
p is nonsingular

px Apx Bu x p 1 Apx p 1 Bu
y Cpx Du

A p 1 Ap A pA p 1
B p 1 B B pB
C Cp C C p 1
D D D D

Sept. 9, 2013 Aerospace Engineering, 21

Definition : Two dynamical systems
x(t ) Ax(t ) Bu (t ) x (t ) A x (t ) B u (t )
&
y (t ) Cx(t ) Du (t ) y (t ) C x (t ) D u (t )
with x(t ) px (t ) are said to be equivalent. The nonsingular
matrix p is called an equivalence transformation.

Theorem: two equivalent dynamical system have the same

transfer function.
G ( s ) C ( sI A ) 1 B D
Cp ( sp 1 Ip p 1 Ap ) 1 p 1 B D
Cp[ p 1 ( sI A) p] 1 p 1 B D
Cpp 1 ( sI A) 1 pp 1 B D
C ( sI A) 1 B D
G ( s)

Sept. 9, 2013 Aerospace Engineering, 22

 Differential equation
Dynamical equation
or Transfer function

d n y (t ) d n1 y (t )
n
1 n 1
n y (t ) u (t )
dt dt
Y ( s) 1
G ( s) n
U ( s) s 1 s n1 2 s n2 n

x (t ) Ax(t ) Bu (t )
y (t ) Cx(t ) Du (t )

Sept. 9, 2013 Aerospace Engineering, 23

 Phase-variable canonical form

x1 y x 1 x2
x 2 x3
x2 y

let x3 y
x n y ( n ) 1 y ( n 1) 2 y ( n 2 ) n y u

n x1 n 1 x2 2 xn 1 1 xn u
xn y ( n 1) y x1
x 1 0 1 0 0 x1 0
x 0 0 1 0 x2 0
2 u


x n n n 1 1 xn 1
x1
x d n y (t ) d n1 y (t )
y 1 0 0 2 1 n 1
n y (t ) u (t )
dt n
dt

xn

Sept. 9, 2013 Aerospace Engineering, 24

 y ( n 1) (0) y ( n 3 ) ( 0) y ( 0)
y ( n 1) xn

u (t ) y (n ) y x2 y x1



y ( n 2 ) xn 1
1

2

n

Sept. 9, 2013 Aerospace Engineering, 25

 xn (0) x1 (0)
s s

1 x n 1 1 1 1
u 1 x n x x 1 x2 1 y
s s n2 s s
x1
1
2

3

n 1

n
x 1 (t ) x2 (t ) sX 1 ( s ) x1 (0) X 2 ( s)
1 x1 (0)
X 1 ( s) s X 2 ( s)
s

Sept. 9, 2013 Aerospace Engineering, 26

Example : d3y d 2 y dy
3
5 2 2y u
dt dt dt

x1 y x 1 0 1 0 x1 0
x 0 x 0 u
let x2 x 1 y 2 0 1 2
x3 x 2 y x 3 2 1 5 x3 1

x3 (0) x 2 ( 0) x1 (0)
s s s

1 1 1
x2 x1
u 1 x 3
s
x3
s s 1 y

5
1
2

Sept. 9, 2013 Aerospace Engineering, 27

 controllable canonical form

y
Y ( s) 1s n 1 2 s n 2 n v
G ( s) n
U ( s) s 1s n 1 2 s n 2 n
v
u
v 1
n v n 1v n 1 2 v n 2 n u
u s 1s n 1 2 s n 2 n

y
1s n 1 2 s n 2 n y 1v n 1 2 v n 2 n
v

Sept. 9, 2013 Aerospace Engineering, 28

 d d
1
dt dt

d
n 1
dt
u (t ) v (n ) v x2 v x1
n


1

2

n

Sept. 9, 2013 Aerospace Engineering, 29

 x1c v x 1c x2c
x2 c v x 2c x3c
let x3c v
x nc n x1c n 1 x2 c 1 xnc u
xnc v ( n 1) y n x1c n 1 x2c 1 xnc

x 1c 0 1 0 0 x1c 0
x 0 0 1 0 x2 c 0
2c u


x nc n n 1 1 xnc 1
x1c
x
Y ( s) 1s n 1 2 s n 2 n
y n n 1 1 2 c G ( s) n
U ( s) s 1s n 1 2 s n 2 n

xnc

Sept. 9, 2013 Aerospace Engineering, 30

 1
2

n 1

xnc x( n 1) c
u (t ) y (n )
x1c y
x2 c
n +

2

n

Sept. 9, 2013 Aerospace Engineering, 31

Example :

4s 3 25s 2 45s 34 s 2 5s 2
G( s) 3 2 3
2s 12s 20s 16
2
2 s 12s 2 20s 16

Answer 2

x 1 0 1 0 x1 0
x 0 0 1 x 0 u (t )
2
2

x 3 8 10 6 x3 1
x1

y 1 2.5 0.5 x2 2u (t )
x3

Sept. 9, 2013 Aerospace Engineering, 32

 observable canonical form

Y (s) 1s n 1 2 s n 2 n
G( s) n 0
U (s) s 1s 2 s n
n 1 n2

1s 1 2 s 2 n s n
0
1 1s 2 s n s
1 2 n

y 1s 1 y 2 s 2 y n s n y 1s 1u 2 s 2u n s nu 0u
x 1o n xno nu
x 2o x1o n 1 xno n 1u
x 3o x2 o n 2 xno n 2u

x nc x( n 1) o 1 xno 1u
y xno nu
Sept. 9, 2013 Aerospace Engineering, 33
 u (t )

n n 1 1 0

y (t )
y (n ) x1o x2 o xno
+ + + +

1 2 n

Sept. 9, 2013 Aerospace Engineering, 34

 Y (s) 1s n 1 2 s n 2 n
G( s) n 0
U ( s) s 1s 2 s n
n 1 n2

x 1o 0 0 0 n x1o n
x 1 0 0 n 1 x2 o n 1
2o u
1

x no 0 0 1 xno 1
x1c
x
y 0 0 1 2 c 0u


xnc

Sept. 9, 2013 Aerospace Engineering, 35

Example :

4s 3 25s 2 45s 34 s 2 5s 2
G( s) 3 2 3
2s 12s 20s 16
2
2 s 12s 2 20s 16

x 1 0 0 8 x1 1
x 1 0 10 x 2.5u (t )
2 2
x 3 0 1 6 x3 0.5
x1
y 0 0 1 x2 2u (t )
x3
Remark T
A0 Ac
T
B0 Cc
T
C0 Bc
Sept. 9, 2013 Aerospace Engineering, 36
 Diagonal & Jordan canonical form

e1 e2 en
G(s)
s 1 s 2 s n

x 1 1 0 0 0 x1 b1
x 0 2 0 0 x2 b2
2 u
0
ei bi ci
x n 0 0 n xn bn
x1
x
y c1 c2 cn 2


xn
Sept. 9, 2013 Aerospace Engineering, 37
 b1 + c1

u (t ) b2 + c2
y (t )

2 +

bn + cn

Sept. 9, 2013 Aerospace Engineering, 38

 x 1 1 1 0 x1 0
x 0 1 1 0 x 0
2 2
x 3 0 0 1 x3 1
x 2 1 x 0
4 4 u
x 5 0 2 x5 1
x 6 0 3 x6 1



e11 e12 e13 e21 e22 e3
G ( s)
( s 1 ) 3 ( s 1 ) 2 ( s 1 ) ( s 2 ) 2 ( s 2 ) s 3 x1
x Jordan block
2
x3
x
y e11 e12 e13 e21 e22 e3 4
x5
x6



Sept. 9, 2013 Aerospace Engineering, 39
Example 4
1 0 1 0.5 0.5
G ( s)
( s 1) ( s 1) ( s 1) s s 2
3 2

x 1 1 1 0 x1 0
x 0 1 1 0 x 0
2 2
x 3 0 0 1 x3 1
x 0 x 4 1 u
4
x 5 0 2 x5 1



x1
x
2
x3
y 1 0 1 0.5 0.5 x4

x5

Sept. 9, 2013
Aerospace Engineering, 40

 Time Response and
State Transition Matrix
Necessary mathematical knowledge:
1. State Transition Matrix
2. Modal decomposition --Diagonalization

3. Cayley-Hamilton Theorem

Sept. 9, 2013 Aerospace Engineering, 41

Time response of the state variables

d
x(t ) Ax(t ) Bu (t )
dt
y (t ) Cx (t ) Du (t )

The behavior of x(t) et y(t) :

1. Homogeneous solution of x(t)
2. Non-homogeneous (particular) solution of x(t)

Sept. 9, 2013 Aerospace Engineering, 42

 Homogeneous solution
x (t ) Ax(t )
x(t ) x(t0 )e A( t t0 )
x(t ) Xe ? At

x(t0 ) e At0 x(0)

At 0
x(0) e x(t0 )
At At 0 A( t t0 )
x(t ) e e x(t0 ) e x(t0 )
Sept. 9, 2013 Aerospace Engineering, 43
 Homogeneous solution
x (t ) Ax(t ) 1 1
x(t ) L [( sI A) ]x(0)
sX ( s ) x(0) AX ( s )
e At x(0)
X ( s ) ( sI A) 1 x(0)
State transition matrix

(t ) e At L1[( sI A) 1 ]
x(t0 ) e At0 x(0)
At0
x(0) e x(t0 )
At At0 A( t t0 )
x(t ) e e x(t0 ) e x(t0 ) (t t0 ) x(t0 )
Sept. 9, 2013 Aerospace Engineering, 44

Properties 1 1
(t ) e L [( sI A) ]
At

1. (0) I
1
2. (t ) (t )
3. x(0) (t ) x(t )
4. (t 2 t1 ) (t1 t0 ) (t 2 t0 )
k
5. (t ) (kt )

Sept. 9, 2013 Aerospace Engineering, 45

 Non-homogeneous solution
d
x(t ) Ax(t ) Bu (t )
dt
y (t ) Cx (t ) Du (t )

sX ( s ) x(0) AX ( s ) BU ( s )
( sI A) X ( s ) x(0) BU ( s )
1 1
X ( s ) ( sI A) x(0) ( sI A) BU ( s )
x(t ) L1[(sI A) 1 ]x(0) L1[( sI A) 1 BU ( s )]
t
x(t ) (t ) x(0) (t ) Bu ( )d Convolution
0

Homogeneous
Sept. 9, 2013 Aerospace Engineering, 46
 t
x(t ) (t ) x(0) (t ) Bu ( )d
0
t
x(t ) (t t0 ) x(t0 ) (t ) Bu ( )d
t0
t
y (t ) C (t t0 ) x(t0 ) C (t ) Bu ( )d Du (t )
t0

Zero-input response Zero-state response

Sept. 9, 2013 Aerospace Engineering, 47

Example 1 x 1 0 1 x1 0
x 2 3 x 1u (t )
2 2
x(0) 0 0
T
let

2e t
e 2 t
e 1 e 2t
(t ) L1[(sI A) 1 ] e At t 2t
2 e 2 e e t 2e 2t

t
x(t ) (t ) x(0) (t ) Bu ( )d
0

x1 2e t e 2t
1 3
Ans: x 2 2 L1[( sI A) 1 BU ( s )]
2 2e 2e 2t
t

Sept. 9, 2013 Aerospace Engineering, 48

 x 1 0 1 x1 0
x 2 3 x 1u (t )
x 2 ( 0)
x1 (0)
s
2 2 s

x ( 0) 0 0
T
let
u 1 s 1 x2 s 1 x1 1 y

3
Using Maisons gain formula 2

1 3s 1 2 s 2
s 1 (1 3s 1 ) s 2 s 2
x1 ( s ) x1 (0) x2 (0) U (s)

2s 2 s 1 s 1
x2 ( s ) x1 (0) x2 (0) U ( s)

Sept. 9, 2013 Aerospace Engineering, 49

 How to find State transition matrix

1 1
(t ) e L [( sI A) ]
At

Methode 1: (t ) L1[( sI A) 1 ]

Methode 2: (t ) e At

Methode 3: Cayley-Hamilton Theorem

Sept. 9, 2013 Aerospace Engineering, 50

 1 1
Methode 1: (t ) L [( sI A) ]
x 1 0 1 0 x1 0 0
x 0 4 3 x 1 0 u1
2 2 u
x 3 1 1 2 x3 0 1 2
x1
y1 (t ) 1 0 0
y (t ) 0 0 1 x2
2 x
3

adj ( sI A)
( sI A) 1
sI A
s 2 6 s 11 s 2 3
1
3 s 2
2 3 s
s ( s 4)( s 2) 3 3s
s4 s 1 s 4 s
2

Sept. 9, 2013 Aerospace Engineering, 51

Method 2: Diagonalization

Example 4.5 x 1 1 0 0 x1 1
x 0 2 0 x 1u diagonal matrix
2 2
x 3 0 0 3 x3 1
x1
y 6 6 1 x2

x3

e t 0 0

(t ) e At 0 e 2 t 0
0 0 e 3t

Sept. 9, 2013 Aerospace Engineering, 52

Diagonalization via Coordinate Transformation

Plant: x Ax Bu A R n n
y Cx Du

Eigenvalue of A: i , satisfying Av i i vi , i 1, ,n
Assume that all the eigenvalues of A are distinct, i.e. 1 2 3 n

Then eigenvectors, v1, v2 , ,vn are independent.

Coordinate transformation matrix T [v1, v2 , ,vn ]

T 1 AT is a diagonal matrix.

Sept. 9, 2013 Aerospace Engineering, 53

 1 0
2
1
T AT


0 n

e At Te tT 1

e1t 0

e2t
where e t

nt
0 e

Sept. 9, 2013 Aerospace Engineering, 54

New T -1 x
coordinate:
T 1 AT
A T 1 AT
1 1 0 0 0 1 B T 1 B

2 0 2 0 0 2 C CT
(4.1)
0

n 0 0 n n

i (t ) ii (t ), i 1, ,n
Solution of (4.1):
i (t ) ei ti (0), i 1, ,n
x (t ) T (t ) v1e1t 1 (0) v2e2t 2 (0) vn ent n (0), (0) T x (0)
1

The above expansion of x(t) is called modal decomposition.

Hence, system asy. stable all the eigenvales of A lie in LHP

Sept. 9, 2013 Aerospace Engineering, 55

Example i distinct
2 1 1
x x , x ( 0 ) 2 1 1, 2 3
1 2
Find eigenvector
1 2 1 v11 v11 1
(1 I A)v1 0
1 1 2 v12 v12 1

3 2 1 v21 v1 1
(2 I A)v2 0
1 3 2 v22 v2 1

1 1 1 1 1 T 1 AT
1 0
T v1 v2
1
T 0 3
1 1 2 1 1

T 1 x

Sept. 9, 2013 Aerospace Engineering, 56

 1 0
(4.2)
0 3

1 11 1 (0)
(0) T x (0), (0)
1

3
2 2 2 ( 0 )

Solution of (4.1):

x (t ) T (t ) v1e1t 1 (0) v2e2t 2 (0)

3
1 2
(0) T x (0) (0)
1

2
3 t 1 3t
3 t 1 1 3t 1 2 e 2 e
x(t ) e e 3 1
2 1 2
1
e e
t 3 t

2 2

Sept. 9, 2013 Aerospace Engineering, 57

In the case of A matrix is phase-variable form and
1 2 3 n

1 1 1
Vandermonde

P v1 v2 vn 1 2 n matrix
for phase-variable
n 1 n 1 form

1 n 1
2 n

1
2
1
P AP
3

4
t 1
e Pe P
At

Sept. 9, 2013 Aerospace Engineering, 58

Example: i distinct
Repeated eigenvalue case, i.e. i is not distinct
1 0 1 1 0 1
A 0 1 0 I A 0 1 0 ( 1)( 1)( 2)
0 0 2 0 0 2

1 2 0 0 1 v1

(1 I A)V1 0 0 0 v2 0
depend
0 0 1 v3

v1 1 v1 0
0v1 0v2 v3 0 v2 0 0v1 0v2 v3 0 v2 1
v3 0 v3 0
V1 V2
Sept. 9, 2013 Aerospace Engineering, 59
3 2 1 0 1 v1
(3 I A)V3 0 1 0 v2 0
0 0 0 v3

v1 1
v1 0v2 v3 0 v2 0
v3 1

1 0 1 1 0 0
P V1 V2 V3 0 1 0 P 1 AP 0 1 0
0 0 1 0 0 2

Sept. 9, 2013 Aerospace Engineering, 60

Case 3: i distinct Jordan form
1 2 3

P v1 v2 v3 P AP Jordan
1
form
Generalized eigenvectors

(1 I A)v1 0 1 1
(1 I A)v2 v1 1
P AP A 1 1
(1 I A)v3 v2 1
e 1t
te 1t t2
2e 1t
A t 1t
e e 1t te
e 1t

Sept. 9, 2013 Aerospace Engineering, 61
Example:
3 1 3 1
A ( 2) 2
1 1 1 1

1 1 v11 v11 1
(1 I A)V1 0
1 1 v12 v12 1
1 1 v21 1 v21 1
(1 I A)V2
1 1 v22 1 v22 0

1 1 2 1
P V1 V2
1
P AP A
1 0 0 2
e 2 t
A t te 2t A t 1
e 2t
e Pe P
At

e
Sept. 9, 2013 Aerospace Engineering, 62
Method 3: Cayley-Hamilton Theorem

Theorem: Every square matrix satisfies its char. equation .

Given a square matrix A, A R nn . Let f() be the char. polynomial of A.

Char. Equation:
f ( ) n an 1n 1 a1 a0 0
By Caley-Hamilton Theorem

f ( A) An an 1 An 1 a1 A a0 I 0

Sept. 9, 2013 Aerospace Engineering, 63

 An an 1 An 1 a1 A a0 I 0
An an 1 An 1 a1 A a0 I
An 1 an 1 An a1 A2 a0 A
an 1 ( an 1 An 1 a1 A a0 I ) a1 A2 a0 A

any f ( A) k0 I k1 A k 2 A2 k n An

f ( A) 0 I 1 A 2 A n 1 A
2 n 1

n 1

k 0
k A k

Sept. 9, 2013 Aerospace Engineering, 64

Example: 1 2
A 100
? A
0 1
let f ( A) A100 0 I 1 A

1 2
( 1)( 2) 0 , 1 1, 2 2
0 2

f (1 ) 1
100
0 11 1100
0 2 2100
f ( 2 ) 2
100
0 12 2 100
1 2100 1

1 0 1 2 1 2101
2
f ( A) A100 (2 2 )
100
(2 1)
100

0 1 0 1 0 1

Sept. 9, 2013 Aerospace Engineering, 65

Example: 3 1
e ?
At
A
2 0
3 1
0 , 1 1, 2 2
2

f (1) e t 0 11 0 1 0 2e t e 2t
f (2) e 2t 0 12 0 1 2 1 e 2 t e t

t 2t1 0 2t t 3 1
e 2e e
At
( e e )
0 1 2 0
2e 2 t e t e 2t e t
2t t 2t t
2e 2e e 2e

Sept. 9, 2013 Aerospace Engineering, 66

Sept. 9, 2013 Aerospace Engineering, 67
Sept. 9, 2013 Aerospace Engineering, 68
Sept. 9, 2013 Aerospace Engineering, 69
Sept. 9, 2013 Aerospace Engineering, 70
Sept. 9, 2013 Aerospace Engineering, 71
Sept. 9, 2013 Aerospace Engineering, 72
Example:
2 0 1 1
x Ax bu 4 1 4 x 1u
2 0 1 0
y 1 0 1 x

1 1 0
0,1,1 v1 4, v 2 0, 1
2 1 0

Sept. 9, 2013 Aerospace Engineering, 73