www.elsevier.com/locate/fss
Decoupled fuzzy controller design with single-input fuzzy logic
Shi-Yuan Chen, Fang-Ming Yu, Hung-Yuan Chung
Department of Electrical Engineering, National Central University, Chungli, Taiwan 320, ROC
Received 27 April 2000; received in revised form 11 June 2001; accepted 19 June 2001
Abstract
A decoupled fuzzy controller design with single-input fuzzy logic is proposed. We will utilize the approach of the
single-input fuzzy logic to explore the fourth-order nonlinear systems. In addition, ve fuzzy rules are given to control a
class of fourth-order nonlinear systems. Using this approach, the system can achieve asymptotic stability and the response
of system will converge faster than that of previous reports. Two simulation studies of a cartpole system and a ballbeam
system are presented to demonstrate the eectiveness of the method. c 2002 Elsevier Science B.V. All rights reserved.
Keywords: Fuzzy control; Sliding mode control; Signed distance
1. Introduction
Fuzzy logic controllers (FLCs) have been proven
to be a powerful tool since the work of Mamdani
[12] was proposed in 1974. The fuzzy control algo-
rithmcan be regarded as model-free control algorithms
in contrast to a conventional feedback control algo-
rithm. However, some issues still exist in the control of
complex systems using fuzzy logic controllers, for
example, (1) No general stability analysis tools can
be applied to FLCs. (2) The large amount of fuzzy
rules for a high-order system make the analysis com-
plex. (3) The design parameters of membership func-
tions aect the performance of the fuzzy system and
suitable membership functions may be obtained via a
1 +c
2
1
: (4)
Without loss of generality, Eq. (4) can be rewritten
as follows:
d =
| x +c
1
x|
1 +c
2
1
: (5)
The signed distance d
s
is dened for an arbitrary point
B(x; x) as follows:
d
s
= sgn(s)
| x +c
1
x|
1 +c
2
1
=
x +c
1
x
1 +c
2
1
=
s
1 +c
2
1
; (6)
where
sgn(s) =
1 for s 0;
1 for s 0:
(7)
For the second-order system (1), a switching line is
chosen as
s = c
1
x
1
+x
2
: (8)
By taking the time derivative of both sides of (8), we
can obtain
s =c
1
x
1
+ x
2
=c
1
x
2
+f(x) +b(x)u +d: (9)
Then, multiplying both sides of the above equation by
s gives
s s = sc
1
x
2
+sf(x) +sb(x)u +sd: (10)
Here, we assume that b(x)0. In (9), it is seen that
s increases as u increases and vice versa. Eq. (10)
provides the information that if s0, the decreasing u
will make s s decrease and that if s0, the increasing
u will make s s decrease. Now, we choose a Lyapunov
function
V =
1
2
d
2
s
: (11)
Then
V = d
s
d
s
=
s s
1 +c
2
1
: (12)
Hence, it is seen that if s0, then d
s
0, decreasing
u will make s s decrease so that
V0 and that if s0,
338 S.-Y. Chen et al. / Fuzzy Sets and Systems 129 (2002) 335342
then d
s
0, increasing u will make s s decrease so that
d
s
= K
1
d
s
(t); (14)
u(t) = K
2
U: (15)
4. Design of decoupled fuzzy logic controller
In this section, the idea of the signed distance of
fuzzy logic control is used in Section 3. In Eq. (2),
we rst dene one switching line as
s
1
= c
1
(x
1
z) +x
2
(16)
and another switching line as
s
2
= c
2
x
3
+x
4
: (17)
The control objective is to drive the system state to
the original equilibrium point. The switching line vari-
ables s
1
and s
2
are reduced to zeros gradually at the
same time by an intermediate variable z.
In Eq. (16), z is a value transferred from s
2
, it has
a value proportional to s
2
and has the range proper to
x
1
. Eq. (16) denotes that the control objective of u
1
is
changed from x
1
=0, x
2
=0 to x
1
=z, x
2
=0.
Because the controller u =u
1
is used to govern the
whole system, the bound of x
1
can be guaranteed by
letting
|z| 6Z
u
; 0 Z
u
1; (18)
where Z
u
is the upper bound of abs(z). Eq. (18) im-
plies that the maximum absolute value of x
1
will be
limited.
Summarizing what we have mentioned above, z can
be dened as
z = sat(s
2
=
z
)Z
u
; 0 Z
u
1; (19)
where
z
is the boundary layer of s
2
to smooth z,
z
transfers s
2
to the proper range of x
1
, and the denition
of sat() function is
sat() =
sgn() if || 1;
if || 1:
(20)
Notice that z is a decaying oscillation signal because
Z
u
is a factor less than one.
Remark 2. Consider Eq. (16). If s
1
=0, then x
1
=z,
x
2
=0. Since z is a value transferred from s
2
, when
s
2
0, then z 0 and x
1
0. From Eq. (17), if
the condition s
1
0, the control objective can be
achieved.
Remark 3. Although SDFLC may be extended to the
order of 2
n
(n =1; 2; 3; : : :) system, it cannot be applied
in higher-order system as in [11]. In a similar way, It
can carry over to the third-order system.
The two-level decoupled SDFLC is proposed and
is shown in Fig. 4.
S.-Y. Chen et al. / Fuzzy Sets and Systems 129 (2002) 335342 339
Fig. 2. Fuzzy variable of triangular type.
Fig. 3. The block diagram of the SDFLC.
Fig. 4. The block of the decoupled SDFLC.
5. Computer simulations
In this section, we shall demonstrate that the decou-
pled SDFLC is applicable to both the cartpole system
and the ballbeam system [11] to verify the theoreti-
cal development.
5.1. Inverted pendulum
The structure of an inverted pendulum is illustrated
in Fig. 5 and its dynamic is described below:
x
1
=x
2
;
x
2
=
m
t
g sin x
1
m
p
Lsin x
1
cos x
1
x
2
2
+cos x
1
u
L(
4
3
m
t
m
p
cos
2
x
1
)
+d;
x
3
=x
4
;
x
4
=
4
3
m
p
Lx
2
2
sin x
1
+m
p
g sin x
1
cos x
1
4
3
m
t
m
p
cos
2
x
1
+
4
3(
4
3
m
t
m
p
cos
2
x
1
)
u +d; (21)
where x
1
= is the angle of the pole with respect to the
vertical axis; x
2
=
the angle velocity of the pole with
340 S.-Y. Chen et al. / Fuzzy Sets and Systems 129 (2002) 335342
Fig. 5. Structure of an inverted pendulum.
respect to the vertical axis; x
3
=x the position of the
cart; x
4
= x the velocity of the cart; and m
t
=m
c
+m
p
.
In what follows, we dene the following variables:
s
1
= c
1
( z) +
= c
1
(x
1
z) +x
2
; (22)
s
2
= c
2
x + x = c
2
x
3
+x
4
(23)
and
z = sat(s
2
=
z
)Z
u
; 0 Z
u
1: (24)
In the simulation, the following specications are
used:
m
p
= 0:05 kg; m
c
= 1 kg; L = 0:5 m;
g = 9:8 m=s
2
; c
1
= 5; c
2
= 0:5;
z
= 15; Z
u
= 0:9425; |d| 60:0873;
K
1
= 1; K
2
= 40
and initial values are
= 60
;
= 0; x = 0; x = 0:
Figs. 68 shows the simulation result. It is found that
the pole and the cart can be stabilized to the equilib-
rium point.
Fig. 6. Angle evolution of the pole. (), z (- - - -).
Fig. 7. Position evolution of the cart.
5.2. Ballbeam system
Consider a ballbeam system as depicted in Fig. 9.
The mathematical expression of this system can be
written as
x
1
= x
2
; x
2
= u +d; x
3
= x
4
;
x
4
= B(x
3
x
2
2
g sin x
1
);
(25)
where x
1
= is the angle of the beam with respect to
the vertical axis; x
2
=
the angle velocity of the beam
with respect to the vertical axis; x
3
=r the position
of the ball; x
4
= r the velocity of the ball; B=MR
2
=
S.-Y. Chen et al. / Fuzzy Sets and Systems 129 (2002) 335342 341
Fig. 8. Control output of Example 1.
Fig. 9. Structure of a ballbeam system.
(J
b
+ MR
2
); J
b
the moment of inertia of the ball; M
the mass of the ball; R the radius of the ball; g the
acceleration of gravity.
The objective is to keep the ball close to the center of
the beamand the beamclose to the horizontal position.
Dene
s
1
= c
1
(x
1
z) +x
2
; (26)
s
2
= c
2
x
3
+x
4
(27)
and
z = sat(s
2
=
z
)Z
u
; 0 Z
u
1: (28)
In the simulation, the following parameters are used:
B = 0:7143; J
b
= 2 10
6
; M = 0:05 kg;
R = 0:01 m; g = 9:8 m=s
2
Fig. 10. Angle evolution of the beam. (), z(- - - -).
Fig. 11. Position evolution of the ball.
c
1
= 5; c
2
= 0:5;
z
= 5; Z
u
= 0:9425;
|d| 60:08; K
1
= 1; K
2
= 40;
the initial values are
x
1
= = 60
; x
2
=
= 0;
x
3
= r = 0; x
4
= r = 0:
The simulation results are shown in Figs. 1012. It
is found that and r converge to zero, respectively.
From Figs. 6 and 7 to 10 and 11, with the same c
1
=5,
342 S.-Y. Chen et al. / Fuzzy Sets and Systems 129 (2002) 335342
Fig. 12. Control output of Example 2.
c
2
=0:5 as in [5], it is seen that the convergence is
faster than that of [11].
6. Conclusions
The present approach was used to replace the fuzzy
controller of Los approach. Compared with previous
results, only ve fuzzy control rules are needed as
given in Table 1. For a ner control, we can easily add
or modify rules by way of the rule table for a signed
distance fuzzy logic control and the SDFLC. Next,
simulation results show that the pole and the cart can
be stabilized to the equilibrium, and r of the ball
beam system converge to zero, and the performance
is better than that of [11]. Although the SDFLC can
be extended to an nth-order system, however, the de-
coupled method only satises the fourth-order system
like the cartpole system. Hence, the same weak point
of the approach is that when applied to systems higher
than fourth order such as a double-inverted pendulum
system (which needs a three-level control), if the cart
is to be stabilized at the origin as in [11], then the
method described here will fail. If there exists a new
SMC method, which can deal with a higher-order sys-
tem larger than the fourth- and the higher-order system
being not a canonical form, then the SDFLC may be
combined with each other. This may be an important
and interesting topic in future.
Acknowledgements
The authors wish to thank the nancial support
of the National Science Council of the Republic of
Taiwan under Contract NSC89-2213-E008-079.
References
[1] G. Bartolini, A. Ferrara, Multivariable fuzzy sliding mode
control by using a simplex of control vectors, in:
S.G. Tzafestas, A.N. Venetsanopoulos (Eds.), Fuzzy
Reasoning in Information, Decision, and Control Systems,
Kluwer, Amsterdam, The Netherlands, 1994, pp. 307328.
[2] C.L. Chen, P.C. Chen, C.K. Chen, Analysis and design of
fuzzy control system, Fuzzy Sets Systems 57 (2) (1993)
125140.
[3] Byung-Jae Choi, Seong-Woo Kwak, Byung Kook Kim,
Design of a single-input fuzzy logic controller and its
properties, Fuzzy Sets and Systems 106 (1999) 299308.
[4] S. Galicher, L. Foulloy, Fuzzy controllers: synthesis
equivalence, IEEE Trans. Fuzzy Systems 3 (1995) 140148.
[5] J.S. Glower, J. Munighan, Design fuzzy controllers from a
variable structures standpoint, IEEE Trans. Fuzzy Systems 5
(1) (1997) 138144.
[6] Guang-Chyan Hwang, Shih-Chang Lin, A stability approach
to fuzzy control design for nonlinear systems, Fuzzy Sets and
Systems 48 (1992) 279287.
[7] Y.R. Hwang, M. Tomizuka, Fuzzy smoothing algorithms for
variable structure systems, IEEE Trans. Fuzzy Systems 2
(1994) 277284.
[8] Sung-Woo Kim, Ju-Jang Lee, Design of a fuzzy controller
with fuzzy sliding surface, Fuzzy Sets and Systems 71 (1995)
359367.
[9] C.C. Kung, S.C. Lin, Fuzzy controller design: a sliding mode
approach, in: S.G. Tzafestas, A.N. Venetsanopoulos (Eds.),
Fuzzy Reasoning in Information, Decision, and Control
Systems, Kluwer, Amsterdam, The Netherlands, 1994, pp.
277306.
[10] K. Liu, F.L. Lewis, Some issues about fuzzy control, in:
proc. IEEE Conf. Decision Control, San Antonio, TX, Vol.
2, December 1993, pp. 17431748.
[11] Ji-Chang Lo, Ya-Hui Kuo, Decoupled fuzzy sliding-mode
control, IEEE Trans. Fuzzy Systems 6 (3) (1998) 426435.
[12] E.H. Mamdani, Applications of fuzzy algorithms for simple
dynamic plants, Proc. IEE, Vol. 121, 1974, pp. 15851588.