Nonlinear Control

Lecture # 6
Passivity
and
Input-Output Stability

Nonlinear Control Lecture # 6 Passivity and Input-Output Stability

Passivity: Memoryless Functions

y

(a)

(b)

Passive

Passive
y = h(t, u),

Vector case: y = h(t, u),

u
(c)

Not passive
h [0, ]

hT =

h1 , h2 , , hp

power inflow = pi=1 ui yi = uT y

Nonlinear Control Lecture # 6 Passivity and Input-Output Stability

Definition 5.1
y = h(t, u) is
passive if uT y 0
lossless if uT y = 0
input strictly passive if uT y uT (u) for some function
where uT (u) > 0, u 6= 0
output strictly passive if uT y y T (y) for some function
where y T (y) > 0, y 6= 0

Nonlinear Control Lecture # 6 Passivity and Input-Output Stability

Sector Nonlinearity: h belongs to the sector [, ]

(h [, ]) if
u2 uh(t, u) u2
y=u

y= u
y

y
y=u
u

u
y=u

(a)

>0

(b)

<0

Also, h (, ], h [, ), h (, )
Nonlinear Control Lecture # 6 Passivity and Input-Output Stability

u2 uh(t, u) u2 [h(t, u) u][h(t, u) u] 0

Definition 5.2
A memoryless function h(t, u) is said to belong to the sector
[0, ] if uT h(t, u) 0
[K1 , ] if uT [h(t, u) K1 u] 0
[0, K2 ] with K2 = K2T > 0 if hT (t, u)[h(t, u) K2 u] 0
[K1 , K2 ] with K = K2 K1 = K T > 0 if
[h(t, u) K1 u]T [h(t, u) K2 u] 0

Nonlinear Control Lecture # 6 Passivity and Input-Output Stability

A function in the sector [K1 , K2 ] can be transformed into a

function in the sector [0, ] by input feedforward followed by
output feedback
+

K 1

y = h(t, u)

[K1 , K2 ]

K1

Feedback
Feedforward
K 1
[0, I]
[0, ]
[0, K]

Nonlinear Control Lecture # 6 Passivity and Input-Output Stability

Passivity: State Models

Definition 5.3
The system
x = f (x, u),

y = h(x, u)

is passive if there is a continuously differentiable positive

semidefinite function V (x) (the storage function) such that
V
uT y V =
f (x, u), (x, u)
x
Moreover, it is
lossless if uT y = V
input strictly passive if uT y V + uT (u) for some
function such that uT (u) > 0, u 6= 0
output strictly passive if uT y V + y T (y) for some
function such that y T (y) > 0, y 6= 0
strictly passive if uT y V + (x) for some positive
definite function

Nonlinear Control Lecture # 6 Passivity and Input-Output Stability

Example 5.2
x = u,

y=x

V (x) = 12 x2 uy = V
x = u,

y = x + h(u),

Lossless
h [0, ]

V (x) = 12 x2 uy = V + uh(u) Passive

h (0, ] uh(u) > 0 u 6= 0
Input strictly passive
x = h(x) + u,

y = x,

h [0, ]

V (x) = 21 x2 uy = V + yh(y) Passive

h (0, ] Output strictly passive
Nonlinear Control Lecture # 6 Passivity and Input-Output Stability

Example 5.3
x = u,
V (x) =

V (x) = a

h [0, ]

h() d V = h(x)x = yu Lossless

ax = x + u,
Z

y = h(x),

y = h(x),

h [0, ]

h() d V = h(x)(x + u) = yu xh(x)

yu = V + xh(x) Passive
h (0, ] Strictly passive

Nonlinear Control Lecture # 6 Passivity and Input-Output Stability

Example 5.4

x 1 = x2 ,

x 2 = h(x1 ) ax2 + u,

h [1 , ],

V (x) =
=

x1

Z0 x1
0

a > 0,

b > 0,

y = bx2 + u
1 > 0

h() d + 21 xT P x
h() d + 21 (p11 x21 + 2p12 x1 x2 + p22 x22 )

> 0,

p11 > 0,

p11 p22 p212 > 0

Nonlinear Control Lecture # 6 Passivity and Input-Output Stability

uy V

= u(bx2 + u) [h(x1 ) + p11 x1 + p12 x2 ]x2

(p12 x1 + p22 x2 )[h(x1 ) ax2 + u]

Take p22 = 1, p11 = ap12 , and = b to cancel the cross

product terms


uy V bp12 1 14 bp12 x21 + b(a p12 )x22
p12 = ak,

0 < k < min{1, 41/(ab)}

p11 > 0, p11 p22 p212 > 0

bp12 1 41 bp12 > 0, b(a p12 ) > 0

Strictly passive

Nonlinear Control Lecture # 6 Passivity and Input-Output Stability

Positive Real Transfer Functions

Definition 5.4
An m m proper rational transfer function matrix G(s) is
positive real if
poles of all elements of G(s) are in Re[s] 0
for all real for which j is not a pole of any element of
G(s), the matrix G(j) + GT (j) is positive
semidefinite
any pure imaginary pole j of any element of G(s) is a
simple pole and the residue matrix limsj (s j)G(s)
is positive semidefinite Hermitian
G(s) is strictly positive real if G(s ) is positive real for
some > 0
Nonlinear Control Lecture # 6 Passivity and Input-Output Stability

Scalar Case (m = 1):

G(j) + GT (j) = 2Re[G(j)]
Re[G(j)] is an even function of . The second condition of
the definition reduces to
Re[G(j)] 0, [0, )
which holds when the Nyquist plot of of G(j) lies in the
closed right-half complex plane
This is true only if the relative degree of the transfer function
is zero or one

Nonlinear Control Lecture # 6 Passivity and Input-Output Stability

Lemma 5.1
An m m proper rational transfer function matrix G(s) is
strictly positive real if and only if
G(s) is Hurwitz
G(j) + GT (j) > 0, R
G() + GT () > 0 or
lim 2(mq) det[G(j) + GT (j)] > 0

where q = rank[G() + GT ()]

Nonlinear Control Lecture # 6 Passivity and Input-Output Stability

Scalar Case (m = 1): G(s) is strictly positive real if and only if

G(s) is Hurwitz
Re[G(j)] > 0, [0, )
G() > 0 or
lim 2Re[G(j)] > 0

Nonlinear Control Lecture # 6 Passivity and Input-Output Stability

Positive Real Lemma (5.2)

Let
G(s) = C(sI A)1 B + D
where (A, B) is controllable and (A, C) is observable. G(s) is
positive real if and only if there exist matrices P = P T > 0, L,
and W such that
P A + AT P = LT L
P B = C T LT W
W T W = D + DT

Nonlinear Control Lecture # 6 Passivity and Input-Output Stability

KalmanYakubovichPopov Lemma (5.3)

Let
G(s) = C(sI A)1 B + D
where (A, B) is controllable and (A, C) is observable. G(s) is
strictly positive real if and only if there exist matrices
P = P T > 0, L, and W , and a positive constant such that
P A + AT P = LT L P
P B = C T LT W
W T W = D + DT

Nonlinear Control Lecture # 6 Passivity and Input-Output Stability

Lemma 5.4
The linear time-invariant minimal realization
x = Ax + Bu,

y = Cx + Du

with
G(s) = C(sI A)1 B + D
is
passive if G(s) is positive real
strictly passive if G(s) is strictly positive real
Proof
Apply the PR and KYP Lemmas, respectively, and use
V (x) = 12 xT P x as the storage function
Nonlinear Control Lecture # 6 Passivity and Input-Output Stability

Connection with Stability

Lemma 5.5
If the system
x = f (x, u),

y = h(x, u)

is passive with a positive definite storage function V (x), then

the origin of x = f (x, 0) is stable
Proof
uT y

V
f (x, u)
x

Nonlinear Control Lecture # 6 Passivity and Input-Output Stability

V
f (x, 0) 0
x

Lemma 5.6
If the system
x = f (x, u),

y = h(x, u)

is strictly passive, then the origin of x = f (x, 0) is

asymptotically stable. Furthermore, if the storage function is
radially unbounded, the origin will be globally asymptotically
stable
Proof
The storage function V (x) is positive definite
uT y

V
f (x, u) + (x)
x

Nonlinear Control Lecture # 6 Passivity and Input-Output Stability

V
f (x, 0) (x)
x

Definition 5.5
The system
x = f (x, u),

y = h(x, u)

is zero-state observable if no solution of x = f (x, 0) can stay

identically in S = {h(x, 0) = 0}, other than the zero solution
x(t) 0
Linear Systems
x = Ax,

y = Cx

Observability of (A, C) is equivalent to

y(t) = CeAt x(0) 0 x(0) = 0 x(t) 0

Nonlinear Control Lecture # 6 Passivity and Input-Output Stability

Lemma 5.6
If the system
x = f (x, u),

y = h(x, u)

is output strictly passive and zero-state observable, then the

origin of x = f (x, 0) is asymptotically stable. Furthermore, if
the storage function is radially unbounded, the origin will be
globally asymptotically stable
Proof
The storage function V (x) is positive definite
uT y

V
f (x, u) + y T (y)
x

V
f (x, 0) y T (y)
x

V (x(t)) 0 y(t) 0 x(t) 0

Apply the invariance principle
Nonlinear Control Lecture # 6 Passivity and Input-Output Stability

Example 5.8
x 1 = x2 , x 2 = ax31 kx2 + u, y = x2 , a, k > 0
V (x) = 41 ax41 + 12 x22
V = ax31 x2 + x2 (ax31 kx2 + u) = ky 2 + yu
The system is output strictly passive
y(t) 0 x2 (t) 0 ax31 (t) 0 x1 (t) 0
The system is zero-state observable. V is radially unbounded.
Hence, the origin of the unforced system is globally
asymptotically stable
Nonlinear Control Lecture # 6 Passivity and Input-Output Stability

L Stability
Input-Output Models:

y = Hu

u(t) is a piecewise continuous function of t and belongs to a

linear space of signals
The space of bounded functions: supt0 ku(t)k <
The
of square-integrable functions:
R space
T
u (t)u(t) dt <
0
Norm of a signal kuk:
kuk 0 and kuk = 0 u = 0
kauk = akuk for any a > 0
Triangle Inequality: ku1 + u2 k ku1 k + ku2 k
Nonlinear Control Lecture # 6 Passivity and Input-Output Stability

Lp spaces:
L : kukL = sup ku(t)k <
t0

L2 : kukL2 =
Lp : kukLp =
Lm
p :

Z

sZ

uT (t)u(t) dt <

1/p
ku(t)k dt
< , 1 p <
p

Notation
p is the type of p-norm used to define the space
and m is the dimension of u

Nonlinear Control Lecture # 6 Passivity and Input-Output Stability

Extended Space: Le = {u | u L,  [0, )}

u(t), 0 t
u is a truncation of u: u (t) =
0,
t>
Le is a linear space and L Le
Example
u(t) = t,

u (t) =

t, 0 t
0,
t>

u
/ L but u Le

Nonlinear Control Lecture # 6 Passivity and Input-Output Stability

q
Causality: A mapping H : Lm
e Le is causal if the value of
the output (Hu)(t) at any time t depends only on the values
of the input up to time t

(Hu) = (Hu )

Definition 6.1
A scalar continuous function g(r), defined for r [0, a), is a
gain function if it is nondecreasing and g(0) = 0
A class K function is a gain function but not the other way
around. By not requiring the gain function to be strictly
increasing we can have g = 0 or g(r) = sat(r)
Nonlinear Control Lecture # 6 Passivity and Input-Output Stability

Definition 6.2
q
A mapping H : Lm
e Le is L stable if there exist a gain
function g, defined on [0, ), and a nonnegative constant
such that

u Lm
e and [0, )

k(Hu) kL g (ku kL ) + ,

It is finite-gain L stable if there exist nonnegative constants

and such that
k(Hu) kL ku kL + ,

u Lm
e and [0, )

In this case, we say that the system has L gain . The bias
term is included in the definition to allow for systems where
Hu does not vanish at u = 0.
Nonlinear Control Lecture # 6 Passivity and Input-Output Stability

L Stability of State Models

x = f (x, u),

y = h(x, u),

0 = f (0, 0), 0 = h(0, 0)

Case 1: The origin of x = f (x, 0) is exponentially stable

Theorem 6.1
Suppose, kxk r, kuk ru ,
c1 kxk2 V (x) c2 kxk2


V
V
2

c4 kxk
f (x, 0) c3 kxk ,
x
x

kf (x, u) f (x, 0)k Lkuk,

kh(x, u)k 1 kxk + 2 kuk

p
Then, for each x0 with kx0 k r c1 /c2 , the system is
small-signal finite-gain Lp stable for each p [1, ]. It is
finite-gain Lp stable x0 Rn if the assumptions hold
globally [see the textbook for and ]
Nonlinear Control Lecture # 6 Passivity and Input-Output Stability

Example 6.4
x = x x3 + u,

y = tanh x + u

V = 12 x2 V = x(x x3 ) x2
c1 = c2 = 21 , c3 = c4 = 1,

L = 1 = 2 = 1

Finite-gain Lp stable for each x(0) R and each p [1, ]

Example 6.5

x 1 = x2 ,

x 2 = x1 x2 a tanh x1 + u,

y = x1 ,

V (x) = xT P x = p11 x21 + 2p12 x1 x2 + p22 x22

Nonlinear Control Lecture # 6 Passivity and Input-Output Stability

a0

= 2p12 (x21 + ax1 tanh x1 ) + 2(p11 p12 p22 )x1 x2

2ap22 x2 tanh x1 2(p22 p12 )x22
p11 = p12 + p22
p22 = 2p12 = 1

the term x1 x2 is canceled

P is positive definite

V = x21 x22 ax1 tanh x1 2ax2 tanh x1

V kxk2 + 2a|x1 | |x2 | (1 a)kxk2
a < 1 c1 = min (P ), c2 = max (P ), c3 = 1 a, c4 = 2c2
L = 1 = 1,

2 = 0

For each x(0) R2 , p [1, ], the system is finite-gain Lp

stable
= 2[max (P )]2 /[(1 a)min(P )]
Nonlinear Control Lecture # 6 Passivity and Input-Output Stability

Case 2: The origin of x = f (x, 0) is asymptotically stable

Theorem 6.2
Suppose that, for all (x, u), f is locally Lipschitz and h is
continuous and satisfies
kh(x, u)k g1 (kxk) + g2 (kuk) + ,

for some gain functions g1 , g2 . If x = f (x, u) is ISS, then, for

each x(0) Rn , the system
x = f (x, u),

y = h(x, u)

is L stable

Nonlinear Control Lecture # 6 Passivity and Input-Output Stability

Example 6.6
x = x 2x3 + (1 + x2 )u2 ,

y = x2 + u

ISS from Example 4.13

g1 (r) = r 2 ,

g2 (r) = r,
L stable

Nonlinear Control Lecture # 6 Passivity and Input-Output Stability

=0

Example 6.7

x 1 = x31 + x2 ,

x 2 = x1 x32 + u,

y = x1 + x2

V = (x21 + x22 ) V = 2x41 2x42 + 2x2 u

x41 + x42 12 kxk4
V

kxk4 + 2kxk|u|
= (1 )kxk4 kxk4 + 2kxk|u|, 0 < < 1
 1/3
(1 )kxk4 , kxk 2|u|
ISS

g1 (r) =

2r,

g2 = 0,

=0

L stable
Nonlinear Control Lecture # 6 Passivity and Input-Output Stability