Professional Documents
Culture Documents
EPCE6101 Linear Systems Theory: (System Stability)
EPCE6101 Linear Systems Theory: (System Stability)
University
ASTU
Department of EPCE
1
Cont…
Open-loop System Regulator
x Ax Bu, x (t0 ) u Kx
y Cx Du
+ u y
r= 0 Plant
-
K x
Feedback System
x ( A BK ) x Ax , x (t0 )
y (C DK ) x
Cont…
Open-loop System Tracking Controller
x Ax Bu, x (t0 ) u Kx Fr
y Cx Du
r + u y
F Plant
-
K x
Feedback System
x ( A BK ) x BFr Ax Br , x (t0 )
y (C DK ) x DFr
2
Cont…
(Definition) Equilibrium state
Given a system with zero input, if a state xe satisfies the
following condition for t t0, it is called equilibrium at
t= t0.
x Ax , x (t0 ) x e 0 Axe
• External input is
regarded as 0
• A zero vector, i.e. xe= 0,
is the equilibrium of all
systems
Cont…
Examples of equilibrium state
• Stable - spring+mass, RLC circuit
• Neutral – DC motor, hydraulic cylinder
• Unstable - scooter, missile, bicycle
3
Cont…
For example,
0 1 0
x x u , x0
1 2 1
Substituting u= 0 and x= xe in the equation yields
0 0 1 xe1 xe 2 0
x e
0 1 2 xe 2 xe1 2 xe 2 0
Cont…
0.5
0 1 x0
x0 0.8
0.8 0.5 Equilibrium
point
x2(t)
-0.5
-1
-1 -0.5 0 0.5 1
0.5 0
x0
x1(t) x0
0.8 0.8
4
Cont…
Example 1: Find equilibrium states.
0 1 0
x x u , x0
0 2 1
Solution: Substituting u= 0 and x= xe in the above
equation yields
0 0 1 xe1 xe 2 0
x e
0 0 2 xe 2 2 xe 2 0
Thus, xe1= , xe2= 0 or xe
0
This system has infinite equilibrium points including
xe= 0.
Cont…
0
x0 1
0.5
0.8 x0
Equilibrium 0.8
0.5
points
x2(t)
-0.5
0.5-1
0
x0 -1
-0.5 0 0.5 1 x0
0.8 x1(t) 0.8
5
Cont…
Example 2: Find equilibrium states.
0 1 0
x x 1 u , x0
2 1
Solution: Substituting u= 0 and x= xe in the above
equation yields
0 0 1 xe1 xe 2 0
x e
0 2 1 xe 2 2 xe1 xe 2 0
0
Solving , yields xe1= xe2= 0 or xe
0
This system has one equilibrium at xe= 0.
Cont…
0
x0 1 0.5
0.8
x0
Equilibrium 0.8
0.5 points
x2(t)
-0.5
0.5 -1
-1 -0.5 0 0.5 1 0
x0 x1(t) x0
0.8 0.8
6
Cont…
Example 3: Find the equilibrium of a two-tank
system below:
dx1 (t ) A 1
1 2 g[ x1 (t ) x2 (t )] qi (t )
dt S1 S1
dx2 (t ) A1 A2
2 g[ x1 (t ) x2 (t )] 2 gx2 (t ) ]
dt S2 S2
qi (t )
where S1
S2
x1(t)= h1(t) h1(t)
h2(t)
x2(t)= h2(t)
h1(t) h2(t) A2
A1
q1 (t ) qo (t )
Cont…
Solution: Substituting qi(t)= 0 and x= xe in yields
dxe1 (t ) A
0 1 2 g[ xe1 (t ) xe 2 (t )]
dt S1
dxe 2 (t ) A A2
0 1 2 g[ xe1 (t ) xe 2 (t )] 2 gxe 2 (t ) ]
dt S2 S2
7
Cont…
Example 4: Find the equilibrium of a Continuous
Stirred Tank Reactor (CSTR) Process
x2
x1 x 1
Da (1 x1 ) exp( )
1 x2 /
x2
x 2 (1 ) x 2
HDa (1 x1 ) exp( ) u
1 x2 /
Cont…
Substituting into yields
(1 ) 1 x
f ( x2 ) x2 Da (1 x2 ) exp( 2
)0
H H 1 x 2
/
xe1 xe2
• one unstable equilibrium point:
2
8
Cont…
A vector x whose distance from the
equilibrium point xe is less than or
x
equal to .
||x - xe||2
x x
x 1 , xe e1
x2 xe 2
|| x xe ||2 ( x1 xe1 ) 2 ( x2 xe 2 ) 2
( x1 xe1 ) 2 ( x2 xe 2 ) 2 2
Cont…
(Definition) Internal stability
The equilibrium is said to be internally stable if, for
any x0 in the region of ||x0-xe|| < , the trajectory of
x starting from x0 stays in the region of ||x-xe|| <
for t 0.
x Ax, x (t0 )
9
Cont…
Test of internal stability
The E.S. xe is said to be internally stable if the
following conditions are met:
Cont…
1= -2 j 1= -2 j Internal
2 = 0 2= j2 stability
3= -j2
0 0
Re(i) 0
1= -2
2= j0 j 1= j2 j Need
3= j0 2= j2 confirmation
= -j2
3= -j2
4
0 0
10
Cont…
Jordan block j
A ()= 3= 0
A= P-1AP 0
Cont…
For example, given a zero-input system
0 1
x x , x0
0 2
Eigenvalues
0 1 1 0 1
A I
0 2 0 1 0 2
1
Δ( ) A I ( 2) 0
0 2
1= 0, 2= -2
Since Re(1)= 0 and Re(2)< 0, the zero-input system
is internally stable.
11
Cont…
For example, given a zero-input system
0 0
x x , x0
2 0
Eigenvalues
0 0 1 0 0
A I 2
2 0 0 1
0
( ) A I 2 0
2
0 𝑟 2
Re(i) 0, it has multiple eigenvalues
Cont…
Method 1: n= 2, r= 2(multiplicity)
Obtaining (A-I) and n-r gives
0 0 0
A I ,
2 0 2 0
(A I) 1
n-r= 2-2= 0
12
Cont…
Method 2: To obtain A, inserting 0 gives
0 0 0
A I 2 0 ,
2 0
( A I ) 1, ( A I ) 2 1 1
0 0
( A I )2 ,
0 0
( A I ) 0, ( A I ) 2 0 2 r
Cont…
0 0 2 0 0
A I , ( A I ) 0 0
2 0
Selecting a vector x such (A-I)2x 0 but (A-I)x 0
and letting x2= x gives
1
x2 x
0
From x2
0 0 1 0
x1 ( A I ) x2
2 0 0 2
13
Cont…
Transformation matrix is
0 1
P [ x1 x2 ]
2 0
From P
1
0 1 0 0 0 1 0 1
Aˆ P 1 AP 2 0 2 0 0 0
2 0
Cont…
0 0
x x , x0
2 0 j
0 𝑟 2
0 1
xˆ xˆ , xˆ 0 0
0 0
0 𝑟 2
14
Cont…
Example 4: Check the stability of a zero-input system.
0 1 0
x 0 0 1 x , x0
0 0 2
= 0(r=2), = -2
Cont…
0 1 0
( A I ) 0 0 0 1 , Unstable
0 0 2
(A I) 2 n r 1
or
0 1 0 Multiple eigenvalue
Aˆ P 1 AP 0 0 0 (= 0) and 2nd Jordan
0 0 2 block in A, Unstable
15
Cont…
10
8 Unstable
variable
6
0
0 1 2 3 4 5
Response with x(t0)= [1 0 0]T
Cont…
Example 5: Check the stability of a zero-input
system.
0 0 1
x 0 0 0 x , x0
0 0 2
16
Cont…
0 0 1
( A I ) 0 0 0 0 ,
0 0 2
(A I) 1 n r Internally stable
or
0 0 0
Aˆ P 1 AP 0 0 0 All 1st-order Jordan block
in A Internally stable
0 0 2
Cont…
1.5
0.5
-1
0 1 2 3 4 5
17
Cont…
(Definition) Asymptotical stability
The equilibrium is said to be asymptotically stable if,
for any x0 in the region of ||x0 - xe|| < , the
trajectory of x starting from x0 always converges to xe
for t 0.
x
xe
Cont…
Test of asymptotical stability
If all the eigenvalues of A have negative real parts, that
is,
Re( ) < 0 (i=1,2,...,n), j
i
18
Cont…
For example,
0 1 0
x x , x0
2 2 0.5
Eigenvalues of A
1
( ) A I 2 2 2 0
2 2
= -1+j, -1-j
Since Re(i) < 0, asymptotically stable.
0 1
x e 0 xe E.S. is xe= 0
2 2
Cont…
For any initial state x0, x(t) always converges to xe= 0,
that is, the origin for t.
x2
x0
0.5
x1
xe= 0
19
Cont…
(Definition) Instability
The equilibrium is said to be unstable if, for any x0 in
the region of ||x0 - xe|| < , the trajectory of x
starting from x0 grows infinitely for t 0.
20
Cont…
For example, given a zero-state system
1 0 1
x x 1 u , x0 0
0 2
The TF X(s)/U(s)= (sI-A)-1B is given by
1 0 1 0 s 1 0
sI A s
0 1 0 2 0 s 2
1
0
1 s 2 0 s 1
( sI A) 1
( s 1)( s 2) 0 s 1 1
0
s 2
Cont…
1 1
s 1 0
1 1 s 1
( s I A) B
0 1 1 1
s 2 s 2
21
Cont…
(Definition) BIBO stability
The state x(t) is said to be BIBO (bounded-input
bounded-output) stable if, for bounded input, that is,
||u(t)|| < k, there exists such that satisfies ||y(t)|| .
Cont…
Example 6: Consider a zero-state system. Is the
system BIBS, BIBO stable?
1 0 0
x x 1 u , x0 0
1 2
y 1 1 x
1 0 1 0 s 1 0
sI A s
0 1 1 2 1 s 2
22
Cont…
1 s 2 0
( sI A) 1 1
( s 1)( s 2) s 1
Thus
1 s 2 0 0
X ( s ) / U ( s ) ( sI A) 1 B 1
( s 1)( s 2) s 1 1
0
1
s 2
Cont…
Obtaining TF Y(s)/U(s)= C(sI-A)-1B+D gives
Y ( s ) / U ( s ) C ( sI A) 1 B D
1 s 2 0 0
=
( s 1)( s 2)
1 1
s 1 1
1
1 0 1
1 1
( s 1)( s 2) s 1 s 2
23
Cont…
Responses x1(t) and x2(t) Response y(t)
0.5 0.5
0.4 0.4
x1(t) y(t) 0.3
0.3
x2(t)
0.2 0.2
0.1 0.1
0 0
0 2 4 6 8 10 0 2 4 6 8 10
t t
Aleksandr Lyapunov
Motion equation (1857-1918)
Russian mathematician
My By ky 0
24
Cont…
Define x1= y, x2= y as state variables
x1 x2
k B
x2 x1 x2
M M
The total energy stored in this system is the sum of
kinetic energy and potential energy.
1 1 1 1
( x1 , x2 ) M ( y ) 2 ky 2 Mx22 kx12
2 2 2 2
kinetic energy potential energy
Cont…
1 1
( x1 , x2 ) Mx22 kx12 0
2 2
Differentiating to see the energy change over time
gives
( x1 , x2 ) Mx2 x2 kx1 x1
k B
Mx2 ( x1 x2 ) kx1 x2 Bx22
M M
If the friction coefficient B is positive,
( x1 , x2 ) Bx22 0
thus, energy eventually dissipates.
25
Cont…
(Theorem) Lyapunov stability
A system
x f ( x , t ), x0
is said to be Lyapunov stable if a scalar Lyapunov
function V(x) satisfying the following conditions
exists.
Cont…
(Definition) Definiteness
For xTAx defined by x 0 and a real symmetric
matrix A,
1. If xTAx >0, A is a positive definite(PD) matrix or
shortly PD
2. If xTAx 0, A is positive semi-definite(PSD)
3. If xTAx <0, A is negative definite(ND)
4. If xTAx 0, A is negative semi-definite (NSD)
5. If xTAx is negative or positive, A is indefinite.
26
Cont…
Test for matrix definiteness
Determinants i= |Ai| (i = 1,2,..,n)
a11 a12
1 a11 , 2 ,..., n | A |
a21 a22
1. If 1>0, 2>0,..., n>0 A is PD
2. If 10, 20,..., n0 A is PSD
3. If 1<0, 2>0, 3<0, 4>0... A is ND
4. If 10, 20, 30, 40... A is NSD
5. Otherwise A is ID
Cont…
Test for matrix definiteness
Eigenvalues of A, i(1 i n)
27
Cont…
(Theorem) Asymptotical stability
A LTI system x Ax , x0 having a point of
equilibrium at xe= 0 is asymptotically stable if, for
any PD matrix Q, the Lyapunov equation
ATP+PA= -Q
has a unique solution P and P is a PD matrix.
(Proof)
Choose V(x)= xTPx as the Lyapunov function, where
P is a PD symmetric matrix. Differentiating V(x)
gives
Cont…
V ( x ) x T Px x T Px ( Ax )T Px xT P ( Ax )
xT ( AT P PA) x
In order for the system to be asymptotically stable, it
must be V(x) < 0 for a PD matrix Q.
V ( x ) x T Qx
Comparing equations and yields
ATP+PA= -Q
Therefore, if a PD matrix P that satisfies the above
equation exists, the system becomes asymptotically
stable.
28
Cont…
Example 7: Show that the system is stable using the
Lyapunov direct method.
0 1
x x AT
P PA Q
-1 -1
Solution: Let P and Q be
p p2 1 0
P PT 1 , QI
p2 p3 0 1
then
0 1 p1 p2 p1 p2 0 1 1 0
1 1 p
2 p3 p2 p3 1 1 0 1
Cont…
p2 p3 p2 p1 p2 1 0
p p
1 2 p2 p3 p3 p2 p3 0 1
29
Cont…
Example 8: Code a program for checking the
stability of the system using the Lyapunov direct
method. 0 1 0
x 0 0 1 x
6 11 6
AT P PA Q I
Program:
>>A= [0 1 0;0 0 1;-6 -11 -6];
>>Q= eye(3,3);
>>P= lyap(A',Q)
>>Lam= eig(P)
Cont…
P=
1.8167 1.1500 0.0833
1.1500 2.0083 0.1500
0.0833 0.1500 0.1083
Lam =
0.0965
0.7610
3.0759
Since i > 0, P is a positive definite matrix and the
system is asymptotically stable.
30
Cont…
Example 9: The following is an unstable open-loop
system. When it has a feedback with u= -[17 9]x,
check the stability of the feedback system using the
Lyapunov direct method.
2 1 0
x x u, x0
0 3 1
Solution: The feedback system is
2 1 0 2 1
x
0 3 x 1 17 9 x 17 6 x, x0
Cont…
Program:
>>A= [2 1;-17 -6]; 2 1
>>Q= eye(2); x x, x0
17 6
>>P= lyap(A',Q)
>>Lam= eig(P) % Eigenvalue
P=
8.2500 1.0000 Since P is a PD matrix, the
1.0000 0.2500 feedback system is
Lam = asymptotically stable.
0.1269
8.3731
31
4. Lyapunov's Method for N.S.
(Definition) Radial unboundness
A function V(x) which is positive definite is called
radially unbounded if V(x)→ +∞ for ||x|| → +∞.
Cont…
Example 10 (Local stability): Consider the zero-
input pendulum system and its state equations.
J (t ) (t ) MgLsin (t ) 0
x1 x2
MgL
x2 sin x1 x2
J J
(1) Obtain the equilibrium point of the system.
(2) Check the stability.
Solution: (1) From the state equations, xe= [n 0]T
(n= 0, 1, 2,…) including xe= 0.
32
Cont…
(2) Consider the following scalar function.
MgL 1
V ( x) (1 cos x1 ) x22
J 2
Note that V is positive semi-definite, that is, V(x) 0.
As a matter of fact, this function represents the sum
of the potential energy and the kinetic energy. Its
time-derivative is easily found to be
MgL
V ( x ) x1 sin x1 x2 x2
J
MgL MgL
x2 sin x1 x2 ( sin x1 x2 ) x22
J J J J
Cont…
V ( x ) x22
J
=0
Equilibrium point (x= xe= [0 0])
• V(x)= 0 for x= 0
• V(x) < 0 for x 0
xe= 0 is locally and asymptotically stable.
33
Cont…
Example 11 (Global stability): Consider the
nonlinear system.
x1 x2 x1 ( x12 x22 )
x2 x1 x2 ( x12 x22 )
(1) Obtain the equilibrium point of the system.
(2) Check the stability.
Solution: (1) From the given state equations,
0 xe 2 xe1 ( xe21 xe22 )
0 xe1 xe 2 ( xe21 xe22 )
Solving these equations gives xe= 0.
Cont…
Let V be the positive definite function
1
V ( x) ( x12 x22 )
2
The derivative of V is
V ( x) x1 x1 x2 x2
x1[ x2 x1 ( x12 x22 )] x2 [ x1 x2 ( x12 x22 )]
( x12 x22 )2 0
which satisfies Condition (2).
The system is a globally asymptotically stable. Note that
the origin is the only equilibrium point of the system.
34
Cont…
1 x 1(t)
x 2(t)
0.5
-0.5
-1
0 10 20 30 40 50
t
Cont…
Example 12 (Instability): Consider the nonlinear
system. Check the stability of xe= 0.
x1 2 x2 x1 ( x12 x24 )
x2 2 x1 x2 ( x12 x24 )
Solution:
Let V be the positive definite function
1 2
V ( x) ( x1 x22 )
2
35
Cont…
The derivative of V is
V ( x) x1 x1 x2 x2
x1[2 x2 x1 ( x12 x24 )] x2 [2 x1 x2 ( x12 x24 )]
( x12 x22 )( x12 x24 ) 0
Cont…
1 x 1(t)
x 2(t)
0.5
-0.5
-1
0 10 20 30 40 50
t
36
Cont…
Example 13 (Stability of linearization system):
Consider the nonlinear system.
x1 ln(2 x22 )
x2 exp( x1 ) exp( x2 )
(1) Find the equilibrium points.
(2) Check their stability.
Solution: (1) From the state equations, there are two
equilibrium points:
Cont…
The Jacobian of the right hand side is
f1 f1
x x 0 2 x2
f 1 2
x22 2
x f 2 f 2 x
e 1 e x2
1x x 2
where f ln(2 x 2 )
1 2
f 2 exp( x1 ) exp( x2 )
Its matrices in x1* and x2* are
f 0 2 f 0 2
A1 A2
x x x1* e e x x x2* 1 / e 1 / e
37
Cont…
The eigenvalues of A1 are
1 1
e j 8e e2
2 2
that are conjugate complex numbers with negative real
parts. Therefore we observe stable spiral around the
equilibrium point x1* . The eigenvalues to A2 are
1 1
( 8e 1 1), ( 8e 1 1)
2e 2e
*
that are one postive and one negative. Therefore, x2 is a
saddle point and is unstable.
Cont…
2
2 1
0
1
-1
0 -2
-3
-1 -4
-2 -1 0 1 2 -5
-6 -4 -2 0 2 4
x1
x1
38
Cont…
Example 14: Consider a system given by
x1 x2
x2 2 x1 3x2 u d
d is the disturbance with d= 0.1sin(x1) and |d(t)| 0.1.
The sliding surface is defined as
s cx1 x2
The sliding mode control law is chosen
u 2 x1 (3 c) x2 sgn( s)
Check the stability of the closed-loop system.
Cont…
Define the Lyapunov function as
1
V s2
2
The derivatives of V and s give x1 x2
x2 2 x1 3x2 u d
V ss
39
Cont…
ss s( sgn( s ) 0.1sin( x1 )) | s | ( | 0.1sin( x1 ) |)
| s | ( 0.1)
x 1 (t)
x 2 (t)
0
0.5
-0.5
0 -1
0 2 4 6 8 10 0 2 4 6 8 10
t t
2 2
Switching function
1.5 0
u(t)
1
-2
0.5
-4
0
0 2 4 6 8 10 0 2 4 6 8 10
t(sec) t
Q&A
40