0% found this document useful (0 votes)
112 views9 pages

Stability PDF

The document discusses stability of linear time-invariant (LTI) systems. It defines three types of stability: 1) Bounded input-bounded output (BIBO) stability, which requires a bounded input to correspond to a bounded output. A system is BIBO stable if its impulse response is absolutely integrable. 2) Internal stability, which requires all poles of the system to be in the open left half of the complex plane. Internal stability implies BIBO stability for controllable and observable systems. 3) Lyapunov stability, which uses a Lyapunov function to show that an equilibrium is stable or asymptotically stable. The document proves Lyapunov stability for
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
112 views9 pages

Stability PDF

The document discusses stability of linear time-invariant (LTI) systems. It defines three types of stability: 1) Bounded input-bounded output (BIBO) stability, which requires a bounded input to correspond to a bounded output. A system is BIBO stable if its impulse response is absolutely integrable. 2) Internal stability, which requires all poles of the system to be in the open left half of the complex plane. Internal stability implies BIBO stability for controllable and observable systems. 3) Lyapunov stability, which uses a Lyapunov function to show that an equilibrium is stable or asymptotically stable. The document proves Lyapunov stability for
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 9

1 Stability

1.1 Bounded Input-Bounded Output (BIBO) Stability


Definition: A system y = Hu is BIBO stable if for any bounded input u(t)
corresponds a bounded output y(t).

In general, the input u(t) and the output y(t) are bounded in the sense of a
signal norm! A scalar signal u(t) is bounded if

Mu < : ku(t)k = sup |u(t)| < Mu .


t0

Definition: A scalar function h(t) is absolutely integrable if Mh < such


that Z
|h( )| d < Mh .
0

Theorem: The SISO linear system with impulse response h(t) is BIBO stable if
and only if h(t) is absolutely integrable.

Theorem: The MIMO linear system with impulse response matrix


H(t) = (Hij (t)) is BIBO stable if and only if hij (t) is absolutely integrable for
all i, j.

MAE 280A 1 Maurcio de Oliveira


Corollary: The MIMO linear system with a rational and proper transfer matrix
H(s) = N (s)/d(s) is BIBO stable if and only if all poles of H(s), i.e., the roots
of d(s), are in the open left-half of the complex plane.
Proof: The impulse response hij (t) can be obtained from Hij (s) as

1 1 Nij (s)
hij (t) = L {Hij (s)} = L .
d(s)
Expanding Hij (s) in partial fractions we have
( m k ) m X ki
i
1
XX i
X
hij (t) = L j
= tj1 ei t
i=1 j=1
(s i ) i=1 j=1

where i denotes the ith root of d(s) with multiplicity ki . Therefore, hij (t) is
absolutely integrable if and only if the factors

tj1 ei t , j = 1, . . . , ki

are absolutely integrable. That is, iff R(i ) < 0.

MAE 280A 2 Maurcio de Oliveira


Proof of Theorem (SISO): Sufficiency: For linear systems
Z
y(t) = h( )u(t ) d
0

and for SISO systems


Z

|y(t)| = h( )u(t ) d ,
Z 0
|h( )u(t )| d,
0
Z
|h( )||u(t )| d.
0

Therefore, since u(t) is bounded and h(t) is absolutely integrable


Z
|y(t)| |h( )||u(t )| d,
0 Z

Mu |h( )| d,
0
M u Mh

which implies y(t) is bounded.


Necessity: If h(t) is not absolutely integrable then there exists t such that
Z t
|h( )| d = .
0

Now for the particular input


(
+1, h(t) 0
u(t t) =
1, h(t) < 0

we have that
Z t Z t
y(t) = h( )u(t ) d = |h( )| d =
0 0

which implies y(t) is not bounded.

MAE 280A 3 Maurcio de Oliveira


1.2 Internal Stability
The LTI system
x(t) = Ax(t) + Bu(t),
(1)
y(t) = Cx(t) + Du(t).
is BIBO stable iff H(s) = C(sI A)1 B + D has all poles on the open
left-half of the complex plane.

The LTI system (1) is internally stable iff all roots of d(s) = det(sI A) are on
the open left-half of the complex plane.

Internal stability = BIBO stability


Internal stability = BIBO stability +
controllability and observability

MAE 280A 4 Maurcio de Oliveira


1.3 Lyapunov Stability
Consider the dynamic system
x = f (x).
Definition: x is an equilibrium point of (1.3) if f (x) = 0.
Definition: Let x = 0 be an equilibrium point of (1.3) and let Rn . It is
1) Stable if for each > 0 there is > 0 such that

kx(0)k < kx(t)k < , t 0.

2) Unstable if not stable.


3) Asymptotically stable if it is stable and > 0 can be chosen such that

kx(0)k < lim x(t) = 0.


t

Theorem: Let x = 0 be an equilibrium point of the dynamic system (1.3) and


x Rn . Let V : R be continuously differentiable and

V (0) = 0, V (x) > 0 x , x 6= 0,

and
V (x) 0, x .
Then x = 0 is stable. Moreover, if

V (x) < 0, x , x 6= 0

then x = 0 is asymptotically stable.

Proof: see H. K. Kalil, Nonlinear Systems, Chapter 3.

MAE 280A 5 Maurcio de Oliveira


1.3.1 Lyapunov Stability for Linear Systems
Consider the LTI system
x = Ax. (2)

Theorem: The following statements about the linear system (2) are equivalent:
1) R(i (A)) < 0.
2) x = 0 is the unique equilibrium point of (2) and it is asymptotically stable.

Proof: Use the same tools to conclude that

x(t) = L (sI A)1 x(0)


so that
lim x(t) = 0
t
for any x(0) bounded iff R(i (A)) < 0. This implies A is nonsingular, which
implies that Ax = 0 x = 0 is the unique equilibrium point.

MAE 280A 6 Maurcio de Oliveira


Theorem: The LTI system (2) is asymptotically stable if and only if, for any
matrix Q > 0, the Lyapunov equation
AT P + P A + Q = 0 (3)
has a unique solution P such that P > 0.

Proof: Sufficiency: Assume there exists P > 0 that solves (3). Then
V (x) = xT P x > 0, for all 0 =
6 x := Rn . Note that
d T
V (x) = x P x,
dt
= xT P x + xT P x,
= xT (AT P + P A)x,
= xT Qx < 0, 0 6= x Rn .
Therefore (2) is asymptotically stable.
Necessity: Assume (2) is asymptotically stable then R(i (A)) < 0. This can be
used to show that the Gramian
Z
T
P = eA t QeAt dt
0

converges to a finite value. Using stability we can also show that


T
lim eA t QeAt = 0.
t

Therefore, as we already know


Z Z
d T d T
AT P + P A = A T eA t QeAt dt + eA t QeAt dt A,
dt dt
Z 0 0
d AT t At
= e Qe dt,
0 dt
T
= lim (eA t QeAt ) Q = Q.
t

To prove that it is positive definite we assume that it is not, so that there exists
z 6= 0 such that (see a similar proof for the Controllability and Observability
Gramians)
Z
T

z P =z eA t QeAt dt = 0 eAt z 0, t 0 z = 0.
0

MAE 280A 7 Maurcio de Oliveira


Finally, to prove that it is unique assume there exists P 6= P satisfying the
Lyapunov equation. Then

AT (P P ) + (P P )A = 0

and
h i h T i h T i
AT t T At T A t At A t At
e A (P P ) + (P P )A e = A e (P P )e + e (P P )e A,
d AT t
= e (P P )eAt = 0, t 0,
dt
which implies that
T
eA t (P P )eAt is constant t 0.

In particular, at t = 0 and t
T
(P P ) = lim eA t (P P )eAt = 0 P = P
t

REMARK #1: The proof works also if Q = C T C 0 and (A, C) is


observable.

REMARK #2: Internal stability Asymptotic stability.

MAE 280A 8 Maurcio de Oliveira


1.4 Example

0 1
x = x
1 1

Lyapunov Equation (Q = I)
T
0 1 p1 p2 p1 p2 0 1 1 0
+ =
1 1 p2 p3 p2 p3 1 1 0 1
is equivalent to the linear system

0 2 0 p1 1
1 1 1 p2 = 0
0 2 2 p3 1
which has the unique solution

p1 3/2
p2 = 1/2 p1 p2 3/2 1/2
= >0
p2 p3 1/2 1
p3 1

MAE 280A 9 Maurcio de Oliveira

You might also like