Scribd
Upload
67 views

I O Stability

1. The document discusses input-output (I/O) stability and Lyapunov stability for continuous and discrete-time systems. 2. For a linear system, I/O stability is equivalent to the integral of the system's impulse response being bounded. 3. Lyapunov stability means that solutions starting near an equilibrium point remain near it, while asymptotic stability additionally requires solutions to converge to the equilibrium over time.

Uploaded by

anuj kumar
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
67 views

I O Stability

1. The document discusses input-output (I/O) stability and Lyapunov stability for continuous and discrete-time systems. 2. For a linear system, I/O stability is equivalent to the integral of the system's impulse response being bounded. 3. Lyapunov stability means that solutions starting near an equilibrium point remain near it, while asymptotic stability additionally requires solutions to converge to the equilibrium over time.

Uploaded by

anuj kumar
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/ 11

ME8281 - Fall 2001, Fall 2002, Spring 2006 (c) Perry Li

I/O Stability

Preliminaries: Vector and function norms

1. Sup norms are used for vectors for simplicity:


kxk∞ = maxi |xi|. Other norms are also okay

2. Induced matrix norms: let A ∈ ℜn×n, (i stands for


induced)

kAxk∞ X
kAki,∞ := sup = max Aji
x∈0 kxk∞ j
i

3. vector function norms: for x : [0, ∞) → ℜn, the


L∞ norm is used, i.e. kx(·)k∞ = supt kx(t)k∞ =
supt |xi(t)|.

M.E. University of Minnesota 84


ME8281 - Fall 2001, Fall 2002, Spring 2006 (c) Perry Li

Definition A system (not necessarily linear) F : u(·) →


y(·) is I/O stable iff there exists k < ∞, s.t. for all
bounded u(·),

ky(·)k∞ ≤ kku(·)k∞

For a linear possibly time varying system, (ignoring


initial conditions) the output is given by:
Z t
y(t) = H(t, τ )u(τ )dτ (8)
−∞

note that this is a linear map between the input function


space and the output function space.

Theorem The linear system (8) is I/O stable iff


Z t 
sup kH(t, τ )kidτ := k < ∞. (9)
t −∞

Because norms are equivalent on finite dimensional


spaces, the norms on vectors (x(t) and y(t)) do not
have to be infinity norms.

M.E. University of Minnesota 85


ME8281 - Fall 2001, Fall 2002, Spring 2006 (c) Perry Li

Specializations

• For system given by:

ẋ(t) = A(t)x(t) + B(t)u(t),


y(t) = C(t)x(t) + D(t)u(t)

with D(t) bounded, I/O stability is equivalent to:


Z t
sup kC(t)Φ(t, τ )B(τ )dτ k =: k < ∞
t −∞

Rt
• For LTI systems, y(t) = −∞
H(t − τ )u(τ )dτ , I/O
stability is given by:
Z ∞
kH(τ )kidτ =: k < ∞.
0

M.E. University of Minnesota 86


ME8281 - Fall 2001, Fall 2002, Spring 2006 (c) Perry Li

• For LTI system ẋ = Ax + Bu, y = Cx + Du, I/O


is equivalent to either 1):
Z ∞
kCeAtBkdt < ∞
0

or 2) the poles of H(s) := C(sI − A)−1B has


negative real parts.
For LTI systems, since poles form subset of {λi(A)},
it therefore suffices to check the eigenvalues of A.
Why ? If poles[H(s)] = {pi} = {λi(A)}, it is
easy to see. For, CeAtB has terms that are of the
form tk eλi t, which is absolutely integrable iff pi has
negative real parts.
If poles[H(s)] is a strict subset of {λi(A)}, then
CeAtB has terms of the form tk epi t. λk 6∈ {poles}
do not appear (see homework problem).

M.E. University of Minnesota 87


ME8281 - Fall 2001, Fall 2002, Spring 2006 (c) Perry Li

Proof (Sufficiency)
Z t

ky(t)k =
H(t, τ )dτ

−∞
Z t
≤ kH(t, τ )kiku(τ )kdτ
−∞
Z t
≤ kH(t, τ )kidτ ku(·)k ≤ kku(·)k.
−∞

(Necessity - sketch)

Rt
1. The function t → −∞
kH(t, τ )dτ k being bounded
is equivalent to:

Z t
∀i, j, t → |Hij (t, τ )|dτ
−∞

(try to prove it using the property of infinity induced


norm).

2. Proof by contraposition. Suppose F : u(·) → y(·)


is I/O stable but (9) is not satisfied. Then, there

M.E. University of Minnesota 88


ME8281 - Fall 2001, Fall 2002, Spring 2006 (c) Perry Li

exists i, j such that


 Z t 
lim gij (t) = |Hij (t, τ )|dτ = ∞.
t→∞ −∞

W.L.O.G., let i = j = 1. Then, there exists


t1, t2, . . . so that g11(tk ) > k.

3. Define a sequence of inputs ul(·), l = 1, 2, 3, · · · by


ul(t) = (ul1(t), 0, · · · )T
(
l sign(h11(t, τ ) t ∈ (−∞, tl]
u1(t) =
0 otherwise

4. For each l, kul(·)k∞ = 1, but


Z tl
l l
ky (·)k ≥ ky (tl)k = |h11(t, τ )|dτ > l
−∞

Thus, as we take l → ∞, ky l(·)k > l → ∞.


Therefore, there can be no k < ∞ so that ky(·)k <
kku(·)k. Q.E.D.

M.E. University of Minnesota 89


ME8281 - Fall 2001, Fall 2002, Spring 2006 (c) Perry Li

Lyapunov (internal) stability

ẋ = f (x, t), f (xe, t) = 0 ∀t. (10)

Definition: The equilibrium, x = xe is stable in the


sense of Lyapunov (i.s.L.) at t0 iff for any ǫ > 0,
there exists δ(ǫ, t0) s.t. if kx(t0) − xek ≤ δ(ǫ, t0),
kx(t) − xek ≤ ǫ for all t ≥ t0.

It is said to be stable if it is stable for each t0 ≥ 0.

Remarks:
• the state can be arbitrarily close to the equilibrium
if initial state is sufficient close.

Definition: The equilibrium, x = xe is asymptotically


stable in the sense of Lyapunov (i.s.L.) at t0 iff

1. xe is stable at t0

2. as t → ∞, x(t) → xe.

M.E. University of Minnesota 90


ME8281 - Fall 2001, Fall 2002, Spring 2006 (c) Perry Li

For linear systems: x(t) = Φ(t, t0)x(t0),

1. x = 0 is stable at arbitrary t0 iff


supt≥t0 kΦ(t, t0)ki = k < ∞

2. x = 0 is asymptotically stable iff Φ(t, t0) → 0 (note


Φ(t, t0) is finite for finite t. Why ?)

Remark:
• x = 0 is stable at t0 iff it is stable at t0 = 0, i.e.
supt≥0 kΦ(t, 0)k := k0.
• x = 0 is asymptotically stable at t0 iff it is
asymptotically stable at t0 = 0, i.e. Φ(t, 0) → 0.
• Hence, if an equilibrium is stable at one initial time,
it is stable at any initial time.

Proof of remark:

kx(t)k = kΦ(t, t0)x0k = kΦ(t, 0)Φ(0, t0)x0k


≤ kΦ(t, 0)kikΦ(0, t0)kikx0k

Notice that for a given t0, kΦ(0, t0)ki is bounded.


Therefore, given ǫ > 0, choose
ǫ
δ(ǫ, t0) = . Q.E.D.
k0kΦ(0, t0)ki

M.E. University of Minnesota 91


ME8281 - Fall 2001, Fall 2002, Spring 2006 (c) Perry Li

Theorem For LTI system, x = 0 is

• stable if each λi(A) has non-positive real part, and


in the case of λi(A) = 0, it is not repeated in a
Jordan form.

• asymptotically stable if and only if each λi(A) has


negative real part.

Proof This is because Φ(t, 0) = exp(At) =


T exp(tΛ)T −1 where Λ is in a Jordan form with
diagonal elements being the eigenvalues. Now exp(tΛ)
consists of terms of the form tk eλit (where k 6= 0 for
Jordan form). If Re(λi) < 0, then each of these terms
→ 0 as t → ∞. If λi = jω (zero real part) and
k = 0, then the exp(tλi) is sinusoidal. If k 6= 0, then
tk eλit has terms of the form tk sin(ωt+φ) which grows
unboundedly.

M.E. University of Minnesota 92


ME8281 - Fall 2001, Fall 2002, Spring 2006 (c) Perry Li

Example: The Jordan form with repeated zero


eigenvalue:  
0 1
A=
0 0
 
1 t
has a transition matrix of exp(tA) = which is
0 1
unbounded (thus x = 0 is unstable).

However, the semi-simple


 
0 0
A=
0 0

has a repeated zero eigenvalue but its transition matrix


is I, which is bounded. Thus, x = 0 is stable.

If  
−2 0
A=
0 −1
 
exp(−2t) 0
the transition matrix is exp(tA) = .
0 exp(−t)
Thus x = 0 is asymptotically stable.

M.E. University of Minnesota 93


ME8281 - Fall 2001, Fall 2002, Spring 2006 (c) Perry Li

Discrete time systems

Equivalent results exist for discrete time systems except


that the pole / eigenvalue locations must be considered
w.r.t. to the unit circle (instead of the left half plane)
in the complex plane. E.g.

• For x(k + 1) = Ax(k), x = 0 is x = 0 is stable if


|λi(A)| ≤ 1, and with no repeated eigenvalues on
the unit circle.

• x = 0 is asymptotically stable iff it is exponentially


stable iff |λi(A)| < 1

M.E. University of Minnesota 94

You might also like