Observer-Based Adaptive Fuzzy Control For SISO Nonlinear Systems Via Backstepping

6th International Conference on Electrical Engineering 11- 13 October 2010
Observer-Based Adaptive Fuzzy Control for

SISO Nonlinear Systems via Backstepping
F. Kara*, R. Abdessemed** and F. Khaber***
*Department of Electrical Engineering, Batna University, Email: [email protected]
**Department of Electrical Engineering, Batna University, Email: [email protected]
***Department of Electrical Engineering, Setif University, Email: [email protected]
Abstract-- In this paper, the observer-based adaptive
fuzzy controllers are developed for a class of SISO
uncertain nonlinear systems with unmeasured states.
Using fuzzy logic systems to approximate the
unknown nonlinear functions, a fuzzy adaptive
observer is introduced for state estimation as well as
system identification. Under the frame work of the
Backstepping design, fuzzy adaptive output feedback
control is constructed recursively. It is proven that the
proposed fuzzy adaptive control approach guarantees the
semi-global boundedness property for all the signals and
the tracking error to a small neighborhood of the origin.
The effectiveness of the proposed approach is verified
through simulation example.
Keywords: Fuzzy control; Adaptive control; Nonlinear
Systems; State observer; Backstepping; Stability.
I. INTRODUCTION
Fuzzy systems adaptive control methodologies is
receiving considerable attention, emerging as
promising approaches for controlling highly uncertain
and nonlinear dynamical systems. Generally, the
adaptive fuzzy control approaches can have good
performance. However, such approaches are based on
the assumption that the state variables of the system
are available for feedback [1].
Recently, fuzzy adaptive backstepping control
schemes can provide a systematic frame work for the
design of tracking or regulation strategies, in which the
fuzzy logic systems (FLS) are used to approximate the
unknown nonlinear functions, and an adaptive fuzzy
controller is constructed recursively[2,3,4]. However,
the existing fuzzy adaptive backstepping control
methods are all based on the assumption that the states
of the systems are directly measured. As pointed out in
[5,6,7], in practice, state variables are often
unmeasured for many nonlinear systems. Therefore,
the problem of fuzzy adaptive backstepping control for
nonlinear systems with state unmeasured is very
important in both theory and real world applications.
Recently, the authors in [5,6,7,8,9] have developed
fuzzy adaptive output feedback control approaches
based on the state observers for uncertain SISO
nonlinear systems. In [4,5,6,7], FLS are utilized to
approximate the unknown functions, the error
observers are designed. In [8], a kind of fuzzy adaptive
output feedback controller is first developed for a class
of uncertain nonlinear systems with no match
condition. Using FLS to approximate the unknown
Sixth International Conference on Electrical Engineering CEE 2010
nonlinear functions, a fuzzy adaptive observer is

introduced for state estimation as well as system
identification. Combining backstepping design with
FLS, fuzzy adaptive output feedback control controller
is constructed recursively, and the adjusting parameter
vectors are derived on the Lyapunov functions
In this paper, our goal is to develop a method for
designing an observer-based adaptive fuzzy controller
for a class of unknown nonlinear systems under the
constraint that only the system output is available for
measurement. Furthermore, for the nonlinear systems
with unknown nonlinearities which are not restricted
to the system output only, an observer-based outputfeedback control law and update law to tune on-line
the weighting factors of the adaptive fuzzy controller
are derived. The output-feedback control law and
update law provide robust stability for the closed-loop
system, and guarantee that all signals involved are
bounded [6].
The paper is organized as follows. In Section II, we
describe the plant dynamics in terms of formulate the
problem and the suggested the observer-based adaptive
fuzzy controller and stability. In Section III, simulation
results are presented to confirm the effectiveness and
applicability of the proposed method. Finally,
conclusions are included in Section IV.
II. PROBLEM FORMULATION AND FUZZY SYSTEMS
A. Nonlinear control problem
Consider a class of the following SISO nth-order
nonlinear system of the form:
x&1 x2=+ f1 ( x1 ) + d1
x& = x + f ( x , x ) + d
3
2
1
2
2
2
M
&
xn -1 = xn + f n -1 ( x1 , K , xn -1 ) + dn -1
x&n = u + f n ( x1 , K , xn ) + dn
y=x
(1)
where X = ( x1 , x2 ,K , xn ) R n is the system state

vector, u R is the control input, y R is the output of
T
system,
fi ( X i )( i = 1,2,Kn) are
unknown
smooth
functions and di ( i= 1, 2,Kn) are the bounded

disturbance uncertainties
of the system. Let
T
X i = ( x1 , x2 ,K , xi ) and it is assumed that X i ( i 2)
are unmeasured. Rewriting (1) in the following form:
7
X& = AX + Ky + Bi f i ( X i ) + d i + Bu
i =1
(2)
where
-k
A= M
- k
, =K
0 K 0
M
k
, =B
0

M , B
1
q i* = arg min sup

fi X i \ qi - fi ( X i \ qi ) ,
qiWi
XiUi1 , X iUi 2
1 i n
(9)
and K is chosen such that A is a strict Hurwitz matrix.

Thus, given a Q>0, there exists a P>0 satisfying
where W i , U i1 and U i 2 are compact regions for q i , X i

and X i , respectively. Also the FLS minimum
approximation errors e i and approximation errors
d i are defined as
(3)
-2Q
Let yr be a reference signal and contain finite
derivatives up to the nth order. Define the output
tracking error as z1 = y - yr .
e = f ( X \ q ) - f X \ q * ,
i
i
i
i
i
i
i
d i = fi ( X i \ qi ) - fi X i \ q i
Denote wi = e i + d i and d i' = d i + d i , i = 1, 2,..., n.
B. Fuzzy logic systems
Assumption 1. There exist known constants d i 0 and ei 0

, i = 1, 2, . . . , n, such that wi e i 0 and d ' d i'0 . Since
=I
0 K 1 K 0
AT P + PA
=
The knowledgebase for FLS comprises a collection

of fuzzy ifthen rules of the following form:
R ( ) : IF x1 is F1l ,and,..., and, xn is Fnl , THEN y is G l ,
l
(4)
l = 1, 2,..., N
where X = ( x1 , x2 ,K, xn ) and y are the FLS input and

T
l
output, respectively. Fuzzy sets Fi and Gl , associated
with the fuzzy functions mF l ( X i ) and mG ( y ) ,
l
respectively and N is the rules number, and

yl = max yR mG ( y ) .
l
Define q T =
[ y , y ,K, y ] = [q ,q ,K,q ] and

1
j ( X ) = j 1 ( X ) , K , j N ( X ) with the fuzzy basis

T
function j l ( x) given by
jl (X ) =
(5)
m l ( X i )
Fi
Then FLS can be rewritten as

l =1
i =1
y ( x) = q T j ( X )
(6)
Lemma1[1]. Let f (x) be a continuous function defined

on a compact set. Then for any constant e > 0 , there
exists an FLS (6) such as
sup f ( x ) - q j ( x ) e
(7)
xW
C. Fuzzy adaptive control design

In this section, it is assumed that the states of
system (1) are not available for feedback, in this
situation, a state observer should be established to
estimate the states, and then fuzzy adaptive output
feedback control scheme is investigated. Due to this
approximation capability, we can assume that the
nonlinear terms in (1) can be approximated as
f i ( X i \ q i ) = qiT ji ( X i ) , fi X i \ q i = qiT ji X i ,
(8)
1 i n
T
where X = ( x , x ,K, x )
Design fuzzy state observer as

X = AX + Ky + n B f X + d + Bu
i i i
i =1
y = CX
( )
(11)
where C = [1 ... 0 ... 0]
Let e = X - X be observer error, then from (2) and

(11), we have the observer errors equation
e& = Ae + Bi f i ( X i ) - fi X i \ qi + d i
i =1
n
= Ae + Bi [d i + d i ] = Ae + d
(12)
Fi
(10)
i =1
P ni=1 m l ( X i )
)
)
the state variables are not available, state observer

should be designed to estimate the states.
(
(
( )
*
Define the optimal parameter vectors qi as
where d = d 1' ,..., d n

The ensuring systematic design procedure involves,
at each step, augmenting an integrator and the design
of a virtual control law to stabilize the augmented
system, until the actual control u appears at the nth
step. The detailed design procedures of fuzzy adaptive
output feedback controller are described in the
following steps:
Step 1. Define the tracking error for the system as
(13)
zi = y - yr
Expressing x2 in terms of its estimate as x2 = x 2 + e2 ,
we obtain
z&1 = x&1 - y& r = x2 + f1 ( x1 ) + d1 - y& r
= x2 + f1 ( x1 ) + d1 - y&r + e2
(14)
= x2 + q j ( x1 ) - y& r + e2 + q% j ( x1 ) + wi
where q% = q * - q
T
1
T
1
Taking x2 as a virtual control, and define

z2 = x2 - a1 ( x1 , q1 , y, yr1 ) - y& r
Then we have
z&1 = z2 + a1 ( x1 ,q1 , y, yr ) + q1T j ( x1 ) + e2 + q%1Tj ( x1 ) + wi
Consider the following Lyapunov function:
V1 =
1 2
1 %T %
z1 +
q1 q1
2
2g 1
(15)
(16)
(17)
where g 1 > 0 is a design constant.

From (12) and (16), the time derivative of V1 is
&
V& = 1 e&T Pe + 1 eT Pe& + z z& + 1 q% T q%
1
g1
1 1
&
V&2 = V&1 + z 2 z 2 + g11 q%2Tq%2
= z1 z2 + a1 ( x1 , q1 , y, yr1 ) + q j ( x1 ) + wi
&
+ e2 z1 + q%1T j ( x1 ) z1 + g11 q%1T q%1T + eT Pd - eT Qe
T
1
By using the inequality 2ab a 2 + b 2 , we have

1
1
2
2
eT Pd + e2 z1 e + Pd + z12
2
2
(18)
(19)
Substituting (19) into (18) yields
(20)
x1
z1
- q1T j1 ( x1 ) - e10 tanh (e10 z1 / k )
2
q&1 = g 1cj1>( x0,1 )kz1>-0,s s(q>1 -0q, 10 ) and q
are
where
a1 = -c1 z1 -
(21)
(22)
design
parameters. In view of (22), use the following nice
property with regard to function tanh (.):
z1w1 - z1e10 tanh ( e10 z1 / k ) 0.2785k = k '
Substituting (21) and (23) into (20) results in
10
V&1 - ( lmin ( Q ) - 1) e - c z + z1 z 2 +
s %T
q1 (q1 - q10 ) + k '
g1
2
1 1
1
2
Pd
Step 2. Differentiating (15) yields

z& = x& - a& ( x , q , y , y ) - &&
y
2
= x3 + H 2 ( x1 , x2 , q1 ,q 2 , y, yr , y& r ) - &&
yr +
-
a1 ' %T
d1 + q 2 j2 X 2 + w2 - d 2'
y
( )
(27)
(28)
a1
e2
y
that z2 w2 - z2e 20 tanh ( e 20 z2 / k ) 0.2785k = k ' ,

from (27) and (28), we have
and
2
1
2
V&2 - ( lmin ( Q ) - 2 ) e - ck z 2 + z2 z3 + d1'2
k =1
1
1
+ d 2'2 + Pd
2
2
2
s %T
q k (q k - q k 0 )
k =1 g
k
k
(29)
+ 2k +
'
Step i. ( 3 i n - 1) :a similar procedure is employed

recursively at each step. By defining
zi = xi - a i -1 x1 ,..., x i -1 , q1 ,..., q i -1 , y , yr ,..., y (i - 2 ) - y (i -1)
we have
z&i = x&i -a&i-1 - y(i) r
= xi+1 + Hi - yr(i) +
(30)
ai-1
a
e2 - i-1 d1' +q%iTji X i + wi - di'
y
y
( )
where
(23)
(24)
( )
where
a
a
H 2 = q j 2 X 2 - 1 x 2 + q1T j 2 ( x 2 ) + k1e1 - 1 q&1
x1
q1
a1
a
x 2 + q1T j1 ( x1 ) - 1 y& r
+ k2 e1 y
yr
Consider the following Lyapunov function:
(25)
V2 = V1 + z 2 z& 2 + g1 q%2T q%2
T
2
a
1
a 2 = - z1 - c2 z2 - z2 - 1 z2 - H2 - e 20 tanh ( e 20 z2 / k )
2
y
&
q 2 = g 2j2 X 2 z1 - s (q2 - q20 )
where q 20 is a design parameter vector. It is noted
are available, the

intermediate control function a1 ( x1,q1, y, yr1 ) and the
adaptation function q1 are chosen as
Taking x 3 as a virtual control and introducing the

variable: z3 = x3 - a 2 ( x1 , x2 , q1 , y, yr , y& r ) - &&
yr
It is noted that variables z and X are available,
2
variables z1 and
( )
choosing intermediate control function a 2 and the

adaptation function q 2 as
V&1 -eT Qe + eT Pd + z1 z2 + 1 + a1 + q1T j ( x1 ) + wi

2
1
2
2
&
T
T
1 %T %
%
+ e + Pd + q1 j ( x1 ) z1 + g1 q1 q1
2
1
z
2
+ Pd + z1 z2 + 1 + a1 + q1T j ( x1 ) + w1
2
2
1 %T
2
&
- ( lmin ( Q ) - 1) e + q1 g 1j ( x1 ) z1 - q1
g1
Since
d
1
2
Pd + q%1T (q1 - q10 )
g1
2
2
a
1
(26)
yr + w2
+ z2 x3 + H 2 + z1 + z 2 - 1 z 2 - &&
2
y
1
1
1
+ q%2T g 2 z 2j 2 X 2 - q&2 + d 1'2 + d 2'2 + k '
2
2
g2
- ( lmin ( Q ) - 2 ) e - c1 z12 +
( )
where g 2 > 0 is a constant. By (23) and (24), and by

using the inequality 2ab a 2 + b 2 , we have the time
derivative of V2
i-1 a
a
H i = q iT ji X i - i-1 x k +1 + q kT j k ( x k ) + ki e1 - 1 q&
k
k =1 x
q1
i -1
i -1 a
a i -1
a1
T
(k)
i-1
e1 -
y - kj
x 2 + q1 j1 ( x1 )
j =1
x j 1 k =1 y r( k -1) r
y
Consider the following Lyapunov function
( )
Vi = Vi -1 +
1 2
1 %T %
qi qi
zi +
2
2g i
(31)
The time derivative of Vi is
a
a
V&i zi xi+1 + H i - y (i ) + i-1 e2 - i-1 d1' + wi - d i'
y
y
r
( ) )
1
+ V&i -1 + q%iT g i ziji X i - q&i
gi
(32)
Again by using the inequality 2ab a 2 + b 2 , we have

a i -1 '
a
1
2
- z id i' - z i
d 1 - z i i -1 e2 zi2 + e +
2
y
y
(33)
2
a i -1 2 1 ' 2 1 ' 2
zi + d 1 + d i
2
2
y
Substituting (32) into (33) results in
1
2
V&i - ( lmin ( Q ) - i ) e + q%iT g i ziji ( X i ) - q&i
gi
2
a
1
+ zi xi+1 + H i + zi-1 + zi - i -1 zi - y r(i ) + wi
2
y
i - 1) '2 i-1 1 '2 1

(
2
'
d1 + d k + + ( i - 1) k Pd
+
k =1 2
2
2
i -1
i -1 s
2
- ck zk + q%kT (q k - qk 0 )
k =1
k =1 g
k
(34)
Introducing the variable

i
z i +1 = x i +1 - a i - y (r )
(35)
Since variables zi and X i are available, choose

intermediate control function a i and adaptation
function qi as
2
1 a
- zi - i-1 zi - Hi -ei0 tanh( ei0 zi /k )
ai =-zi-1 - cz
i i
2 y
(36)
&
q = g j X z -s (q -q )
i
i i
( )
i
i0
where g i > 0 and qi 0 are design constants. By (34)

(36), (34) becomes
i-1
( i - 1) d '2
2
V&i - ( lmin ( Q ) - i ) e - ck zk2 + zi zi +1 +
1
2
k =1
(37)
i -1
i-1
1 '2 1
d %T
2
+ d k + Pd + q k (qk - q k 0 ) + ( i ) k '
2
k =1 2
k =1 g k
Step n. In the final design step, the actual control input
u appears. We consider the over all Lyapunov function
as
1
1 %T %
qn qn
Vn = Vn -1 + z n2 +
2
2g n
(38)
By setting i=n, the control u and adaptation

functions q n are described by
2
a
u = - z n -1 - cn z n - H n - n -1 z n
y
- e n 0 tanh ( e n 0 z n / k ) + yr( n )
q& = g j X z - s (q - q )
n
( )
n
n0
V&n - ( lmin ( Q ) - n ) e - ck z +
1
+ 1 + Pd
2
k =1
n
2
k
( n - 1)
2
'2
10
s
d + q k* - q k 0 + nk '
k =1
k =1 g
k
n
'2
k0
Defining fuzzy basis functions as

( x - 3 + j ) 2
exp - 1
16
, j=1,,5.
j1 j ( x1 ) =
( x1 - 3 + n ) 2
5
n=1 exp 16
( x - 3 + j )2
( x - 3 + j )2
* exp - 2
exp - 1
4
4
j 2 j ( x1 , x2 ) =
2
( x1 - 3 + n )
( x - 3 + n )2
5
* exp - 2
n=1 exp 4
16
The FLS can be expressed in the form

5
q1T j1 ( x1 ) = q1Tjj1 j ( x1 ) ,
j =1
q j 2 ( x1 , x2 ) =
T
2
q
j =1
T
2j
j 2 j ( x1 , x 2 )
Design parameters in controller and in adaptive

laws are chosen as
K = [ k1 , k 2 ]=T
T
[2,1]=
; g1
g 2= 0.1; s= 0.2
- k1 1 - 2 1
A= =
= = =; c1
- k 2 0 -1 0
=c2
5; e 10
e 20
0.1
For the given symmetric positive matrix

Q= diag [ 4,4] , by solving Lyapunov equation (3), we
have the following symmetric positive matrix P:
(39)
P = [ 4 - 4; - 4 12]
The initial conditions are chosen as
(40)
x1 ( 0 ) = 0, x2 ( 0) = -0.2, x1 ( 0)= 0, x2 ( 0 )= 0.3
we can obtain that the time derivative of Vn is

2
( x - 3 + l )2
, l=1,,5.
m F l ( x1 ) = exp - 1
1
16
( x - 3 + l )2
( x - 3 + l )2
* exp - 2
m F l ( x1 , x2 ) = exp - 1
1
4
16
q10 = q20 = [ 0, 0,0;0;0]

(41)
q1 ( 0 ) = [-0.6;0;0.5;0;0.6] , q2 ( 0 ) = [0.4;0;0.1;-0.1;0]
0.6
III. SIMULATION AND RESULT

Consider a system governed by the following form:
x&1 (t ) = x 2 (t ) + x1e -0.5 x1
2
(42)
x& 2 (t ) = u + x1 (t ) sin ( x 2 )
y ( t ) = x1 ( t )
where f1 = x1e -0.5 x and f 2 = x1 sin ( x22 ) are unknown
1
functions. Assume that state x1 and x2 are unmeasured,

the given tracking reference signal is y r = sin(t ) .
Choosing fuzzy member ship functions as
0.4
0.2
-0.2
-0.4
10
15
Seconds
20
25
30
Fig. 1 The trajectories of x1 solidline

and x1 dashdotted
10
matching conditions, but also can deal with a class of

nonlinear systems with unmeasured states and no
matching conditions.
1
0.8
0.6
0.4
IV. CONCLUSION
0.2
0
-0.2
-0.4
-0.6
10
15
Seconds
20
25
30
Fig. 2 The trajectories of x2 solidline

and x2 dashdotted
0.6
0.4
0.2
-0.2
In this paper, the adaptive fuzzy control methods

based on Backstepping design has been proposed for
SISO nonlinear systems. The main contribution of this
paper is that by designing a fuzzy adaptive state
observer, the application of adaptive fuzzy
Backstepping control is extended to a new class of
nonlinear systems with unmeasured states. In addition,
the stability of the closed-loop system has been proved
by using Lyapunov method, i.e., the proposed adaptive
fuzzy control scheme can guarantee the closed-loop
system semi-global stability and the convergence of
the tracking error of the system
to a small
neighborhood of the origin. Simulation results have
been given to illustrate the effectiveness of the
proposed scheme.
-0.4
-0.6
-0.8
V. REFERENCES
0
10
15
Seconds
20
25
30
Fig. 3 The trajectories of x1 solidline and yr dash

dotted
[1] S. Tong, Y. Li Observer-based fuzzy adaptive control

for strict-feedback nonlinear systems, Fuzzy Sets and
Systems, 2008.
0.5
[2] B. Chen, X.P. Liu, S.C. Tong, Fuzzy approximate

disturbance decoupling of MIMO nonlinear systems by
backstepping approach, Fuzzy Sets and Systems 158
(2007) 10971125.
0.4
0.3
0.2
0.1
[3] S.C. Tong, Y.M. Li, Direct adaptive fuzzy

Backstepping control for a class nonlinear systems,
Internat. J. Innovative Comput. Informat. Control 3
(2007) 887896.
0
-0.1
-0.2
-0.3
-0.4
-0.5
10
15
Seconds
20
25
30
Fig. 4 Trajectories of the tracking

errors of system z1 = x1 - y r
[5] A. Boulkroune, M. Tadjine, M. MSaad, M. Farza,

How to design a fuzzy adaptive controller based on
observers for uncertain affine nonlinear systems,
Fuzzy Sets and Systems 159 (2008) 926948.
2.5
2
1.5
[6] Y.G. Leu, W.Y. Wang, Observer-based adaptive

fuzzyneural control for unknown nonlinear dynamical
systems, IEEE Trans. Systems Man Cybernet. 29
(1999) 583591.
1
0.5
0
-0.5
-1
[4] M. Wang, B. Chen, Direct adaptive fuzzy tracking

control for a class of perturbed strict-feedback nonlinear
systems, Fuzzy Sets and Systems 158 (2007) 2655
2670.
10
15
Seconds
20
25
30
Fig. 5 The trajectory of u.
We can conclude that the proposed fuzzy control

approach can guarantee the boundedness of the
signals x1 , x1 , x 2 , x 2 and u. Especially, the tracking error
z1 = x1 - y r converges to as mall neighborhood of the
origin. The proposed fuzzy controller not only control
SISO nonlinear systems with unmeasured states and
[7] S.C. Tong, H.X. Li, W. Wang, Observer-based

adaptive fuzzy control for SISO nonlinear systems,
Fuzzy Sets and Systems 148 (2004) 355376.
[8] F.Y. Hsu, L.C. Fu, A novel adaptive fuzzy variable
structure control for a class of nonlinear uncertain
systems via backstepping, Fuzzy Sets and Systems 22
(2001) 83106.
[9] W.Y. Wang, Y.G. Leu, T.T. Lee, Output-feedback
control of nonlinear systems using direct adaptive
fuzzyneural controller, Fuzzy Sets and Systems 140
(2003) 341358.
11

Observer-Based Adaptive Fuzzy Control For SISO Nonlinear Systems Via Backstepping

Uploaded by

Copyright:

Available Formats

Observer-Based Adaptive Fuzzy Control For SISO Nonlinear Systems Via Backstepping

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Observer-Based Adaptive Fuzzy Control For SISO Nonlinear Systems Via Backstepping

Uploaded by

Copyright:

Available Formats

6th International Conference on Electrical Engineering 11- 13 October 2010

Observer-Based Adaptive Fuzzy Control for

nonlinear functions, a fuzzy adaptive observer is

where X = ( x1 , x2 ,K , xn ) R n is the system state

functions and di ( i= 1, 2,Kn) are the bounded

q i* = arg min sup

and K is chosen such that A is a strict Hurwitz matrix.

where W i , U i1 and U i 2 are compact regions for q i , X i

Denote wi = e i + d i and d i' = d i + d i , i = 1, 2,..., n.

B. Fuzzy logic systems

Assumption 1. There exist known constants d i 0 and ei 0

The knowledgebase for FLS comprises a collection

where X = ( x1 , x2 ,K, xn ) and y are the FLS input and

respectively and N is the rules number, and

[ y , y ,K, y ] = [q ,q ,K,q ] and

j ( X ) = j 1 ( X ) , K , j N ( X ) with the fuzzy basis

Then FLS can be rewritten as

Lemma1[1]. Let f (x) be a continuous function defined

C. Fuzzy adaptive control design

Design fuzzy state observer as

where C = [1 ... 0 ... 0]

Let e = X - X be observer error, then from (2) and

the state variables are not available, state observer

Sixth International Conference on Electrical Engineering CEE 2010

where d = d 1' ,..., d n

Taking x2 as a virtual control, and define

where g 1 > 0 is a design constant.

By using the inequality 2ab a 2 + b 2 , we have

Substituting (19) into (18) yields

Step 2. Differentiating (15) yields

that z2 w2 - z2e 20 tanh ( e 20 z2 / k ) 0.2785k = k ' ,

Step i. ( 3 i n - 1) :a similar procedure is employed

q 2 = g 2j2 X 2 z1 - s (q2 - q20 )

where q 20 is a design parameter vector. It is noted

are available, the

Taking x 3 as a virtual control and introducing the

choosing intermediate control function a 2 and the

V&1 -eT Qe + eT Pd + z1 z2 + 1 + a1 + q1T j ( x1 ) + wi

where g 2 > 0 is a constant. By (23) and (24), and by

Sixth International Conference on Electrical Engineering CEE 2010

The time derivative of Vi is

Again by using the inequality 2ab a 2 + b 2 , we have

i - 1) '2 i-1 1 '2 1

Introducing the variable

Since variables zi and X i are available, choose

where g i > 0 and qi 0 are design constants. By (34)

By setting i=n, the control u and adaptation

Defining fuzzy basis functions as

The FLS can be expressed in the form

Design parameters in controller and in adaptive

For the given symmetric positive matrix

x1 ( 0 ) = 0, x2 ( 0) = -0.2, x1 ( 0)= 0, x2 ( 0 )= 0.3

we can obtain that the time derivative of Vn is

q10 = q20 = [ 0, 0,0;0;0]

III. SIMULATION AND RESULT

functions. Assume that state x1 and x2 are unmeasured,

Sixth International Conference on Electrical Engineering CEE 2010

Fig. 1 The trajectories of x1 solidline

matching conditions, but also can deal with a class of