Applied Nonlinear Control Nguyen Tan Tien - 2002.

5
_________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________

6. Feedback Linearization

Feedback linearization is an approach to nonlinear control ~

This implies that h (t ) → 0 as t → ∞ . From (6.2) and (6.3),
design.
the actual input flow is determined by the nonlinear control
- The central idea of the approach is to algebraically
law
transform a nonlinear system dynamics in to a fully or
partly one, so that the linear control theory can be applied.
- This differs entirely from conventional linearization u (t ) = a 2 gh − A(h) α (h) (6.5)
(such as Jacobian linearization) in that the feedback,
rather than by linear approximations of the dynamics. Note that in the control law (6.5)
- Feedback linearization technique can be view as ways of a 2 gh : used provide the output flow
transforming original system models into equivalent
models of a simpler form. A(h) α (h) : used to rise the fluid level according to
the desired linear dynamics (6.4)
6.1 Intuitive Concepts If the desired level is a known time-varying function hd (t), the
This section describes the basic concepts of feedback ~
equivalent input v can be chosen as v = h& (t ) − α h so as to d
linearization intuitively, using simple examples.
~
still yield h (t ) → 0 when t → ∞ . □
6.1.1 Feedback linearization and the canonical form
Example 6.1: Controlling the fluid level in a tank ⊗ The idea of feedback linearization is to cancel the
Consider the control of the level h of fluid in a tank to a nonlinearities and imposing the desired linear dynamics.
specified level hd. The control input is the flow u into the tank
and the initial value is h0. Feedback linearization can be applied to a class of nonlinear
u system described by the so-called companion form, or
controllability canonical form.

Consider the system in companion form

h
output
flow  x&1   x2 
&   
x
 2 =  x 3 
Fig. 6.1 Fluid level control in a tank (6.6)
 M   M 
   
 x n   f ( x ) + b( x) u 
&
The dynamic model of the tank is
where
h 
d  x : the state vector
dt  ∫
A(h)dh  = u (t ) − a 2 gh

(6.1)
f ( x), b( x) : nonlinear function of the state
o  u : scalar control input
where, A(h) is the cross section of the tank, a is the cross
section of outlet pipe. The dynamics (6.1) can be rewritten as For this system, using the control input of the form

u = (v − f ) / b (6.7)
A(h) h& = u − a 2 gh (6.2)
we can cancel the nonlinearities and obtain the simple input-
If u(t) is chosen as output relation (multiple-integrator form) x ( n) = v . Thus, the
control law v = − k 0 x − k1 x& − K − k n −1 x ( n −1) with the ki chosen
u (t ) = a 2 gh + A(h)v (6.3)
so that the polynomial p n + k n −1 p n −1 + K + k 0 has its roots
with v being an “equivalent input” to be specified, the strictly in the left-half complex plane, lead to exponentially
resulting dynamics is linear h& = v stable dynamics x ( n ) + k n −1 x ( n −1) + K + k 0 x = 0 which
Choosing v as implies that x(t ) → 0 . For tasks involving the tracking of the
~ desired output xd (t), the control law
v = −α h (6.4)
~
with h = h(t ) − hd is the level error, α is a strictly positive v = x d ( n) − k 0 e − k1e& − K − k n −1e ( n −1) (6.8)
constant. Now, the close loop dynamics is
(where e(t ) = x(t ) − x d (t ) is the tracking error) leads to
~
h& + α h = 0 (6.4) exponentially convergent tracking.
___________________________________________________________________________________________________________
Chapter 6 Feedback linearization 26
 Applied Nonlinear Control Nguyen Tan Tien - 2002.5
_________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________

Example 6.2: Feedback linearization of a two-link robot first put the dynamics into the controllability canonical form
Consider the two-link robot as in the Fig. 6.2 before using the above feedback linearization design.

6.1.2 Input-State Linearization

Consider the problem of design the control input u for a
single-input nonlinear system of the form
I2, m2
lc2 x& = f ( x,u )
l2
l1 q 2, τ 2
The technique of input-state linearization solves this problem
into two steps:
lc1
I 1, m 1
- Find a state transformation z = z ( x ) and an input trans-
q1,τ1 formation u = u( x, v ) , so that the nonlinear system
dynamics is transformed into an equivalent linear time-
invariant dynamics, in the familiar form z& = A z + b v .
Fig. 6.2 A two-link robot - Use standard linear technique to design v .

The dynamics of a two-link robot Example: Consider a simple second order system

 H 11 H 12   q&&1  − h q& 2 − h q&1 − h q& 2   q&1   g1  τ 1  x&1 = −2 x1 + a x 2 + sin x1 (6.11a)

   =   &  +   =  
 H 21 H 22  q&&2   h q&1 0  q 2   g 2  τ 2  x& 2 = − x 2 cos x1 + u cos(2 x1 ) (6.11b)
(6.9)
where, Even though linear control design can stabilize the system in a
q = [q1 q 2 ]T : joint angles
small region around the equilibrium point (0,0), it is not
obvious at all what controller can stabilize it in a large region.
τ = [τ 1 τ 2 ]T : joint inputs (torques) A specific difficulty is the nonlinearity in the first equation,
which cannot be directly cancelled by the control input u.
H 11 = m1l c21 + I 1 + m 2 (l12 + l c22 + 2l1l c 2 cos q 2 ) + I 2 Consider the following state transformation
H 12 = H 21 = m2 l1c 2 cos q 2 + m2 lc22 + I 2
z1 = x1 (6.12a)
H 22 = m 2 l c22 + I 2 z 2 = a x 2 + sin x1 (6.12b)
h = m 2 l1l c 2 sin q 2
g1 = m1l c1 g cos q1 + m 2 g[l c 2 cos(q1 + q 2 ) + l1 cos q1 ] which transforms (6.11) into
g 2 = m 2 l c 2 g cos(q1 + q 2 )
z&1 = −2 z1 + z 2 (6.13b)
Control objective: to make the joint position q1 and q 2 z& 2 = −2 z1 cos z1 + cos z1 sin z1 + a u cos(2 z1 ) (6.13b)
follows desired histories q d1 (t ) and q d 2 (t )
The new state equations also have an equilibrium point at (0,0).
Now the nolinearities can be canceled by the control law of
To achieve tracking control tasks, one can use the follow the form
control law
1
τ 1   H 11 H 12   v1  − h q& 2 − h q&1 − h q& 2   q&1   g1  u= (v − cos z1 sin z1 + 2 z1 cos z1 ) (6.14)
 =   =   &  +   a cos(2 z1 )
τ 2   H 21 H 22  v 2   h q&1 0  q 2   g 2 
(6.10) where v is equivalent input to be designed (equivalent in the
sense that determining v amounts to determining u, and vise
where,
versa), leading to a linear input-state relation
v& = q&&d − 2λq~& − λ 2 q~
z&1 = −2 z1
v = [v1 v 2 ]T : the equivalent input
(6.15a)
z& 2 = v (6.15b)
q~ = q − q d : position tracking error
λ : a positive number
Thus,
state
The tracking error satisfies the equation q&~& + 2λq~& + λ 2 q~ = 0 the problem of
transformation
the problem of
and therefore converges to zeros exponentially. stabilizing the original stabilizing the new
(6.12)
nonlinear dynamics dynamics (6.15)
(6.11) using the original input using the new
⊗ When the nonlinear dynamics is not in a controllability transformation
canonical form, one may have to use algebraic transforms to control input u input v
(6.14)
___________________________________________________________________________________________________________
Chapter 6 Feedback linearization 27
 Applied Nonlinear Control Nguyen Tan Tien - 2002.5
_________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________

Now, consider the new dynamics (6.15). It is linear and &y& = ( x 2 + 1) u + f1 ( x ) (6.21)
controllable. Using the well known linear state feedback
control law v = − k1 z1 − k 2 z 2 , one could chose k1 = 2, k 2 = 0 f1 ( x ) = ( x15 + x 3 )( x 3 + cos x 2 ) + ( x 2 + 1) x12 (6.22)
or
Clearly, (6.21) represents an explicit relationship between y
v = −2 z 2 (6.16) and u . If we choose the control input to be in the form

1
resulting in the stable closed-loop dynamics z&1 = −2 z1 + z 2 u= (v − f 1 ) (6.23)
x2 +1
and z& 2 = −2 z 2 . In term of the original state, this control law
corresponds to the original input where v is the new input to be determined, the nonlinearity in
(6.21) is canceled, and we apply a simple linear double-
1 integrator relationship between the output and the new input v,
u= (−2 a x 2 − 2 sin x1 − cos x1 sin x1 + 2 x1 cos x1 )
a cos(2 x1 ) &y& = v . The design of tracking controller for this double-
(6.17) integrator relation is simple using linear technique. For
The original state x is given from z by instance, letting e = y (t ) − y d (t ) be the tracking error, and
choosing the new input v such as
x1 = z1 (6.18a)
x 2 = ( z 2 − sin z1 ) / a (6.18b) v = &y&d − k1e − k 2 e& (6.24)

The closed-loop system under the above control law is where k1 , k 2 are positive constant. The tracking error of the
represented in the block diagram in Fig. 6.3. closed-loop system is given by

0 x &e& + k 2 e& + k1e = 0 (6.25)

v=- k Tz u=u (x,v) x& =f(x,u)

linearization loop which represents an exponentially stable error dynamics.

Therefore, if initially e(0) = e&(0) = 0 , then e(t ) ≡ 0, ∀t ≥ 0 ,
pole-placement loop z i.e., perfect tracking is achieved; otherwise, e(t ) converge to
z=z (x)
zero exponentially.
Fig. 6.3 Input-State Linearization ⊗ Note that:
⊗ To generalize the above method, there are two equations: - The control law is defined anywhere, except at the
- What classes of nonlinear systems can be transformed singularity point such that x 2 = −1 .
into linear systems ? - Full state measurement is necessary in implementing the
- How to find the proper transformations for those which control law.
can ? - The above controller does not guarantee the stability of
internal dynamics.
6.1.3 Input-Ouput Linearization
Consider a tracking control problem with the following system Example 6.3: Internal dynamics
Consider the nonlinear control system
x& = f ( x,u ) (6.19a)
y = h( x ) (6.19b)  x&1   x 23 + u 
&  =   (6.27a)
 x 2   u 
Control objective: to make the output y (t ) track a desired
y = x1 (6.27b)
trajectory y d (t ) while keeping the whole state bounded.
y d (t ) and its time derivatives are assumed to be known and Control objective: to make y track to y d (t )
bounded.
y& = x&1 = x 23 + u ⇒ u = − x 23 − e(t ) + y& d (t ) (6.28)
Consider the third-order system
yields exponential convergence of e to zero.
x&1 = sin x 2 + ( x 2 + 1) x 3 (6.20a)
x& 2 = x15 + x 3 (6.20b) e& + e = 0 (6.29)
x& 3 = x12 +u (6.20c) Apply the same control law to the second dynamic equation,
y = x1 (6.20d) leading to the internal dynamics

To generate a direct relationship between the output and input, x& 2 + x 23 = y& d − e (6.30)
let us differentiate the output y& = x&1 = sin x 2 + ( x 2 + 1) x 3 .
Since y& is still not directly relate to the input u , let us which is non-autonomous and nonlinear. However, in view of
differentiate again. We now obtain the facts that e is guaranteed to be bound by (6.29) and y& d is
___________________________________________________________________________________________________________
Chapter 6 Feedback linearization 28
 Applied Nonlinear Control Nguyen Tan Tien - 2002.5
_________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________

assumed to be bounded, we have y& d (t ) − e ≤ D , where D is

positive constant. Thus we can conclude from (6.30) that
x 2 ≤ D1 / 3 , since x& 2 < 0 when x 2 > D1 / 3 , and x& 2 > 0 when

x 2 < − D1 / 3 . Therefore, (6.28) does represent a satisfactory

tracking control law for the system (6.27), given any trajectory
y d (t ) whose derivative y& d (t ) is bounded.

⊗ Note: if the second state equation in (6.27a) is replaced by

x& 2 = −u , the resulting internal dynamics is unstable.

▲ The internal dynamics of linear systems

⇒ refer the test book
▲ The zero-dynamics
Definition: The zeros-dynamics is defined to be the internal
dynamics of the systems when the system output is kept at
zero by the input.

For instance, for the system (6.27)

 x&1   x 23 + u 
&  =   (6.27a)
 x 2   u 
y = x1 (6.27b)

the out put y = x1 ≡ 0 → y& = x&1 ≡ 0 → u ≡ − x 23 , hence the

zero-dynamics is

x& 2 + x 23 = 0 (6.45)

This zero-dynamics is easily seen to be asymptotically stable

by using Lyapunov function V = x 22 .

⊗ The reason for defining and studying the zero-dynamics is

that we want to find a simpler way of determining the stability
of the internal dynamics.
- In linear systems, the stability of the zero-dynamics
implies the global stability of the internal dynamics.
- In nonlinear systems, if the zero-dynamics is globally
exponentially stable only local stability is guaranteed for
the internal dynamics.

⊗ To summarize, control design based on input-output

linearization can be made in three steps:
- differentiate the output y until the input u appears.
- choose u to cancel the nonlinearities and guarantee
tracking convergence.
- study the stability of the internal dynamics.

6.2 Mathematical Tools

6.3 Input-State Linearization of SISO Systems

6.4 Input-Output Linearization of SISO System

___________________________________________________________________________________________________________
Chapter 6 Feedback linearization 29