Scribd
Upload
47 views

Lect 35

The document discusses feedback linearization and sliding mode control for tracking control of nonlinear systems. It presents the normal form for feedback linearization of a SISO system with relative degree ρ. Tracking control is achieved by defining a tracking error and designing the control law to stabilize the error dynamics. Sliding mode control is also discussed, where a sliding surface is defined in terms of the tracking error and its derivatives, and a discontinuous control law is designed to drive the system state onto and maintain it on the sliding surface.

Uploaded by

win alfalah
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
47 views

Lect 35

The document discusses feedback linearization and sliding mode control for tracking control of nonlinear systems. It presents the normal form for feedback linearization of a SISO system with relative degree ρ. Tracking control is achieved by defining a tracking error and designing the control law to stabilize the error dynamics. Sliding mode control is also discussed, where a sliding surface is defined in terms of the tracking error and its derivatives, and a discontinuous control law is designed to drive the system state onto and maintain it on the sliding surface.

Uploaded by

win alfalah
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 11

Nonlinear Systems and Control

Lecture # 35

Tracking

Feedback Linearization &


Sliding Mode Control

p. 1/1
SISO relative-degree system:

x = f (x) + g(x)u, y = h(x)

f (0) = 0, h(0) = 0
1
Lg Li1
f h(x) = 0, for 1 i 1, Lg Lf h(x) 6= 0
Normal form:
= f0 (, )
i = i+1 , 1i1
= Lf h(x) + Lg L1
f h(x)u
y = 1

f0 (0, 0) = 0

p. 2/1
Reference signal r(t)
r(t) and its derivatives up to r () (t) are bounded for all
t 0 and the th derivative r () (t) is a piecewise
continuous function of t;
the signals r ,. . . ,r () are available on-line.

Goal: lim [y(t) r(t)] = 0


t


r 1 r

R= ..
e= ..
. , . =R
r (1) r (1)

p. 3/1
= f0 (, e + R)
h i
1
e = Ac e + Bc Lf h(x) + Lg Lf h(x)u r ()

0 1 0 ... 0 0

0 0 1 ... 0
0



.. ... .. ..
Ac = . . , Bc =
.

..

. 0
1 0
0 ... ... 0 0 1

1 h

i
u= 1 Lf h(x) + r () + v
Lg Lf h(x)
e = Ac e + Bc v

p. 4/1
v = Ke e = (Ac Bc K) e
| {z }
H urwitz

lim e(t) = 0 lim [y(t) r(t)] = 0


t t

e(t) is bounded (t) = e(t) + R(t) is bounded


What about (t)?
= f0 (, )
Local Tracking (small k(0)k, ke(0)k, kR(t)k):

Minimum Phase The origin of = f0 (, 0) is


asymptotically stable
is bounded for sufficiently small
k(0)k, ke(0)k, and kR(t)k
p. 5/1
Global Tracking (large k(0)k, ke(0)k, kR(t)k):

What condition on = f0 (, ) is needed?

Example 13.21

x1 = x2 , x2 = a sin x1 bx2 + cu, y = x1

e1 = x1 r, e2 = x2 r
e1 = e2 , e2 = a sin x1 bx2 + cu r
1
u= [a sin x1 + bx2 + r k1 e1 k2 e2 ]
c
e1 = e2 , e2 = k1 e1 k2 e2
See simulation in the textbook
p. 6/1
Sliding Mode Control

x = f (x) + g(x)[u + (t, x, u)], y = h(x)

Lg h(x) = = Lg Lf2 h(x) = 0, Lg L1


f h(x) a > 0

= f0 (, )
1 = 2
.. ..
. .
1 =
1
= Lf h(x) + Lg Lf h(x)[u + (t, x, u)]
y = 1

e=R

p. 7/1
= f0 (, )
e1 = e2
.. ..
. .
e1 = e
e = Lf h(x) + Lg L1
f h(x)[u + (t, x, u)] r ()
(t)

Sliding surface:

s = (k1 e1 + + k1 e1 ) + e

s(t) 0 e = (k1 e1 + + k1 e1 )

p. 8/1
= f0 (, )
e1 = e2
.. ..
. .
e1 = (k1 e1 + + k1 e1 )

Design k1 to k1 such that the matrix



1

...

is Hurwitz

1
k1 k1

Assumption: The system f0 (, ) is BIBS stable

p. 9/1
1
X
s = (k1 e1 + + k1 e1 ) + e = ki ei + e
i=1

1
X 1
s = ki ei+1 +Lf h(x)+Lg Lf h(x)[u+(t, x, u)]r () (t)
i=1
"1 #
1 X
u= 1 ki ei+1 + Lf h(x) r () (t) + v
Lg Lf h(x) i=1

s = Lg L1
f h(x)v + (t, x, v)

(t, x, v)

1 (x) + 0 |v|, 0 0 < 1
Lg L h(x)
f

p. 10/1
 
s
v = (x) sat , >0

(x)
(x) + 0 , 0 >
(1 0 )
What properties can we prove for this control?

p. 11/1

You might also like