Professional Documents
Culture Documents
Backstepping approach
This process continues so that each state zi is stabilized by the ctitious control zi+1 .
x = fx (x) + gx (x)z1
until
the ultimate control u is designed.
...
Recursive
Design
..
Overview
3 INTEGRATOR BACKSTEPPING
is stabilized so that z1 follows the desired ux control. The control design is
based on the augmented Lyapunov function candidate
3 Integrator Backstepping
Before describing the backstepping procedure for general strict-feedback form dynamical systems, it is convenient to discuss the approach for a smaller class of
1
V1 (x, z1 ) = Vx (x) + (z1 ux (x))2
strict-feedback form systems. These systems connect a
2
series of integrators to the input of a system with a known
The control u1 can be picked to bound
feedback-stabilizing control law, and so the stabilizing
V 1 away from zero.
approach is known as integrator backstepping. With a
small modication, the integrator backstepping approach
3. A control u2 (x, z1 , z2 ) is designed so that the system can be extended to handle all strict-feedback form systems.
z2 = f2 (x, z1 , z2 ) + g2 (x, z1 , z2 )u2 (x, z1 , z2 )
is stabilized so that z2 follows the desired u1 control. The control design is
based on the augmented Lyapunov function candidate
1
V2 (x, z1 , z2 ) = V1 (x, z1 ) + (z2 u1 (x, z1 ))2
2
The control u2 can be picked to bound
V 2 away from zero.
x = fx (|{z}
0 ) + (gx (|{z}
0 ))(|{z}
0 ) = 0 + (gx (0))(0) = 0
control uk1 .
x
u1
The ctitious control uk1 stabilizes zk1 to
z}|{
(i.e., z1 =
The ctitious control uk2 stabilizes zk2 to So the origin (x, z ) = (0, 0) is an equilibrium (i.e., a
1
ctitious control uk3 .
stationary point) of the system. If the system ever reaches
the origin, it will remain there forever after.
...
The ctitious control u2 stabilizes z2 to ctitious control u1 .
The ctitious control u1 stabilizes z1 to ctitious control ux .
x = F (x)
(i.e., x =
z1
where
Vx
V x =
(fx (x) + gx (x)ux (x)) W (x)
x
3.2
Single-integrator Backstepping
where W (x) is a positive-denite function.
That is, we assume that we have already
shown that this existing simpler x subsystem
is stable (in the sense of Lyapunov). Roughly
speaking, this notion of stability means that:
3
We now can change variables from (x, z1 ) to (x, e1 )
by letting e1 z1 ux (x) . So
We seek to stabilize this error system by feedback through the new control v1 . By stabilizing the system at e1 = 0 , the state z1 will track
the desired control ux which will result in stabilizing the inner x subsystem.
From our existing Lyapunov function Vx , we dene
the augmented Lyapunov function candidate
1
V1 (x, e1 ) Vx (x) + e21
2
So
V x (i.e.,
z = u
1
1
which we can re-group to get
z1 = u1
x = F (x)
d Vx
dt
z }| {
e 1
z}|{ Vx
1
W (x)
z
}|
{
Vx
Vx
Vx
V1 =
(fx (x) + gx (x)ux (x)) +
gx (x)e1 +e1 v1 W (x)+
g
x
x
x
To ensure that V 1 W (x) < 0 (i.e., to ensure stability of the supersystem), we pick the
control law
v1 =
k1 > 0
Vx
gx (x) k1 e1
x
3 INTEGRATOR BACKSTEPPING
v1
z(
}|
){
V
V
x
x
V 1 = W (x)+
gx (x)e1 +e1
gx (x) k1 e1 Before discussing the recursive procedure for the general
x
x
multiple-integrator case, it is instructive to study the recursion present in the two-integrator case. That is, consider the dynamical system
e
1
x = fx (x) + gx (x)ux
that has scalar input ux and output states x =
[x1 , x2 , . . . , xn ]T Rn . Assume that
3.4
Many-integrator backstepping
the system.
z[}|]{ z[
}|
{
z}|{
]
[ ]
z1
0
1
integrator backstepping procedure can be applied again. That is, if we feed back states z1 ,
z2 , and x to input u2 according to the control
law
u2 (x, z1 , z2 ) =
{
x = fx (x) + gx (x)z1
z1 = u1
This cascade system matches the form in
Equation (1), and so the single-integrator backstepping procedure leads to the stabilizing control law in Equation (3). That is, if we feed
back states z1 and x to input u1 according to
the control law
V1
u1
g1 (x1 )k2 (z2 u1 (x1 ))+
(f1 (x1 )+g1 (x1 )z2 )
x1
x1
x = fx (x) + gx (x)z1
z1 = z2
z2 = u2
Connect an integrator to input u2 so that the augmented system has input u3 and output states x . The
resulting augmented dynamical system is
x = fx (x) + gx (x)z1
z = z
1
2
=
z
2
3
z3 = u3
which can be re-grouped as the singleintegrator system
f2 (x2 )
g2 (x2 )
x 2
z
}|
{
z}|{
z}|{
x
f
(x)
+
g
(x)z
0
x
x
2
z2
z1 =
+ 0 z3
z2
0
1
z3 = u3
function Lyapunov by
GENERIC BACKSTEPPING
function Lyapunov by ( Vx ,
x = fx (x) + gx (x)z1
z1 = z2
f2 (x2 )
g2 (x2 )
x 2
z[}|]{ z[
}|
{ z}|{
]
[
]
z2 = z3
x 1
f1 (x1 ) + g1 (x1 )z2
0
by
z3
function
Lyapunov
=
+
1
z
zi = zi+1
z3 = u3
..
k2 = zk1
zk1 = zk
{
zk = u
x 2 = f2 (x2 ) + g2 (x2 )z3
function Lyapunov by ( V2 , by stabilized subsystem u2 (x2 ))
and can be feedback stabilized by nding
z3 = u3
the feedback-stabilizing control and Lyapunov
function for the single-integrator (x, z1 ) subSo the re-grouped system has the singlesystem
(i.e., with input z2 and output x ) and
integrator structure of Equation (1), and so the
iterating
out from that inner subsystem until
single-integrator backstepping procedure can
the
ultimate
feedback-stabilizing control u is
be applied again. That is, if we feed back states
known.
At
iteration
i , the equivalent system is
z1 , z2 , z3 , and x to input u3 according to the
control law
x i1
z }| {
x
V2
u2
u3 (x, z1 , z2 , z3 ) =
g2 (x2 )k3 (z3 u2 (x2 ))+
(f
2 (x
z1 z
}|
z}|{
[
]{
[ ]
z2
f
(x
)
+
g
(x
)z
0
i2
i2
i2
i1
i2
with gain k3 > 0 , then the states z1 , z2 ,
+
zi
func.
. =
0
1
i2
i1
zi = ui
function is
1
V3 (x, z1 , z2 , z3 ) = V2 (x2 ) + (z3 u2 (x2 ))2
2
x = fx (x) + gx (x)z1
z1 = z2
z2 = z3
..
.
zi = zi+1
..
zk2 = zk1
zk1 = zk
zk = u
}|
{
z
Vi1
ui
gi1 (xi1 ) ki (zi ui1 (xi1 )) +
ui (x, z1 , z2 , . . . , zi ) =
xi1
xi
with gain ki > 0 . The corresponding Lyapunov function is
1
Vi (xi ) = Vi1 (xi1 ) + (zi ui1 (xi1 ))2
2
By this construction, the ultimate control
u(x, z1 , z2 , . . . , zk ) = uk (xk ) (i.e., ultimate
control is found at nal iteration i = k ).
4 Generic Backstepping
Systems in the special strict-feedback form have a recursive structure similar to the many-integrator system
4.2
Many-step Procedure
4.1
Single-step Procedure
Because this strict-feedback system has a feedbackstabilizing control and a corresponding Lyapunov function, it can be cascaded as part of a larger strict-feedback
system, and this procedure can be repeated to nd the
surrounding feedback-stabilizing control.
where
As in many-integrator backstepping, the single-step procedure can be completed iteratively to stabilize an entire
strict-feedback system. In each step,
x = [x1 , x2 , . . . , xn ]T Rn ,
z1 and u1 are scalars,
For all x and z1 , g1 (x, z1 ) = 0 .
Rather than designing feedback-stabilizing control u1 directly, introduce a new control ua1 (to be designed later)
and use control law
u1 (x, z1 ) =
1
(ua1 f1 (x, z1 ))
g1 (x, z1 )
x = fx (x) + gx (x)z1
Equation (6) is
..
x
=
f
(x)
+
g
(x)z
x
x
1
u1 (x,z1 )
zi = fi (x, z1 , z2 , . . . , zi ) + gi (x, z1 , z2 , . . . , zi )zi+1
}|
{
z
..
(ua1 f1 (x, z1 ))
z1 = f1 (x, z1 ) + g1 (x, z1 )
g1 (x, z1 )
function Lyapunov
x = fx (x) + gx (x)z1
=
f
(x,
z
)
+
g
(x,
z
)z
=
f
(x,
z
,
z
)
+
g
(x,
z1 , z2 )z3
2
2
1 2
2
..
ux
Vx
1 ) .
trol law is
zk = fk (x, z1 , z2 , . . . zk1 , zk ) + gk (x, z1 , z2 , . . . , zk1 , zk )u
and can be feedback stabilized by nding the feedbackstabilizing control and Lyapunov function for the singleintegrator (x, z1 ) subsystem (i.e., with input z2 and output x ) and iterating out from that inner subsystem until
the ultimate feedback-stabilizing control u is known. At
iteration i , the equivalent system is
REFERENCES
[2] Lozano, R.; Brogliato, B. (1992). Adaptive control of robot manipulators with exible joints. IEEE
Transactions on Automatic Control 37 (2): 174181.
doi:10.1109/9.121619.
[3] Khalil, H.K. (2002). Nonlinear Systems (3rd ed.). Upper
Saddle River, NJ: Prentice Hall. ISBN 0-13-067389-7.
x i1
z }| {
fi1 (xi1 )
gi1 (xi1 )
1 z
}|
{ z[ }| ]{
[
]
z
fi2 (xi2 ) + gi2 (xi2 )zi2
0
2
zi
+
. =
g
f
(x
)
i1 (xi )
i1 i
.
zi2
zi1
xi
z
}|
{
ui (x, z1 , z2 , . . . , zi ) =
1
Vi (xi ) = Vi1 (xi1 ) + (zi ui1 (xi1 ))2
2
By this construction,
the ultimate control
u(x, z1 , z2 , . . . , zk ) = uk (xk ) (i.e., ultimate control is found at nal iteration i = k ). Hence, any
strict-feedback system can be feedback stabilized using
a straightforward procedure that can even be automated
(e.g., as part of an adaptive control algorithm).
See also
Nonlinear control
Strict-feedback form
Robust control
Adaptive control
}|
1
ui1
Vi1
gi (xi ) xi1
xi1
References
7.1
Text
7.2
Images
7.3
Content license