Jianwei Cheng
School of Aerospace Science
Beijing institute of technology
Beijing, China
cjw03919@bit.edu.cn
Xiaoxian Yao
School of Aerospace Science
Beijing institute of technology
Beijing, China
yxx11@bit.edu.cn
Zhiyuan Yu
China Academy of Launch Vehicle
Technology
Beijing, China
happytimmyu@yahoo.cn
Abstract—Fuzzy sliding mode controller based on trending law
for electric actuator is proposed in this paper. The sliding mode
is used to enhance the robustness of the system, the trending law
is used to improve the system dynamic characteristics of
reaching the sliding surface, and the fuzzy tuning schemes are
employed to reduce the chattering problem around the sliding
surface. The integral sliding mode controller with exponential
reaching law was developed. And the parameters of the
exponential trending law were configured by fuzzy control rule.
The method is applied to the control of an electric actuator. And
the results of the simulation verify the validity of the proposed
approaches.
I. INTRODUCTION
ncertainty in dynamical non-linear systems which
suffer from external disturbances and parameters
variation is considered as a difficult challenge in designing
the control system. Sliding mode control has been suggested
as an approach for the control of systems with nonlinearities,
uncertain dynamics and bounded input disturbances for the
last two decades and was widely recognized as a potential
approach to this problem. The most distinguished feature of
the sliding mode control technique is its ability to provide fast
error convergence and strong robustness for control systems
in the sense that the closed loop systems are completely
insensitive to nonlinearities, uncertain dynamics, uncertain
system parameters and bounded input disturbances in the
sliding mode.
However, chattering phenomena will take place when
implementing a sliding mode control as there are
discontinuous control actions. Chattering problem is an
obvious drawback of sliding mode in application. The reasons
are as follows. In a real system, the sliding mode variable
structure around the sliding surface accompanied with high
frequency chattering as there are time delay switch, space
delay switch, system inertia, measure error and other factors.
The chattering not only affects the control accuracy, increase
the energy consumptions, but also may excite the
high-frequency un-modeled dynamics parts.
One common way to eliminate this drawback is to
introduce the trend law to sliding mode. This method can not
only improve the system dynamic characters of reaching the
sliding surface but also eliminate high frequency chattering of
the control signal through adjusting the parametersε and k .
Furthermore, ε may bring chattering problems when it’s too
big. Fuzzy tuning schemes are employed in order to solve this
problem effectively. The fuzzy tuning law is proposed by
analyzing the relationship between the switching function and
the parameters of exponential turning law. The parameters
ε and k are the output of the fuzzy controller.
In this paper, an integral sliding mode controller is
developed to minish the electric actuator system error. The
exponential turning law is employed to improve the dynamic
characters of reaching the sliding surface. Moreover, Fuzzy
U
Fuzzy Sliding Mode Controller Based on Trending Law for Electric
Actuator
978-1-4244-6044-1/10/$26.00 ©2010 IEEE
eq su u u= +
control and sliding mode are combined together to eliminate
the chattering problem.
II. SLIDING MODE BASED ON EXPONENTIAL TRENDING LAW
The mathematic model of a typical electric actuator system
is given by Eq. (1) as follows. It is obvious that the system is
a second derivative system.
4
2
( )
( ) v
Ks
U s T s s
δ = +
(1)
We design a sliding mode scheme for the following n
degree-of-freedom linear SISO system
( , )x Ax Bu Df x t= + +� (2)
Where, nx∈ℜ is the state vector of the system, u is the
input of the system, ( , )f x t is the external disturbances of the
system,
0D is the maximal value of the disturbances.
Usually we define the switching function as
1 1 2 2( ) n ns x c x c x c x= + + +" (3)
When 1nc = , and Eq. (3) can been rewrite as
1 1 2 2( ) ns x c x c x x= + + +" (4)
In order to reduce the system error, a new state vector is
introduced, and it is defined as 0 1x x=� , which means
0 1( ) ( )dx t x t t= ∫ (5)
Thus, extending the system to 1n + degree-of-freedom and
the system can be described as
( , )Ex A x bu f x t= + +� (6)
Where, 1nx +∈ℜ .
The switching function is
0 0 1 1 2 2( ) ns x Cx c x c x c x x= = + + + +" (7)
Use Ackermann-formula to choose
, 0,1, 2, , 1ic i n= −"
Then the Ackermann-formula is expressed as
( )T EC e P A= (8)
[ ] 110, ,0,1 , , ,T nE Ee b A b A b −−⎡ ⎤= ⎣ ⎦" " (9) ( )( ) ( )( )0 1 1( ) n nP λ λ λ λ λ λ λ λ λ−= − − − −"
2 2
0 1 2 1
n n
nP P P Pλ λ λ λ−−= + + + + +" (10)
1
0
[1,0, ,0]0
I
E I n
A
A AA ×
⎡ ⎤= =⎢ ⎥⎣ ⎦
"
Where, ( )P λ is the characteristic polynomial of the
system, 0 1 nλ λ λ", ,, are the expectant eigenvalues of the
switch surface.
The control function u is make up of two parts, and is
shown as
(11)
Where, equ is the equivalent control, the function of this
term is to force the system state to slide on the sliding surface,
and su is the switch control part which is used to meet the
needs of compensating the uncertainty, non-linear and other
unideal factors of the system.
Consider Eq. (11) and the exponential turning law together,
we can get a new expression as
( ) ( )1 0( )sgn( )u CB CAx ks D sε−= − + + + (12)
Where, the equivalent control part and the switch control
part can been written as follows.
( ) 1equ CB CAx−= − (13)
( ) [ ]1 0( ) sgn( )su CB ks D sε−= − + + (14)
Where 0D can be ignored, and use sgn( )sε to compensate
it.
III. FUZZY CONTROL
The fuzzy controller use s and s� as system input,ε and k as
system output. The scheme of fuzzy sliding mode controller
can be shown as figure 1.
We define that
1 2n ns k s ksε σ ε σ= − = −� (15)
1ns sα= (16)
2n ns sα=� � (17)
Where, 1 2 1 2, , ,σ σ α α are quantization factors, and they are
all positive.
Moreover, we define the fuzzy sets of input ,n ns s� and
output nε , k as: { }3 2 1 0 1 2 3, , , , , ,A A A A A A A− − − −� � � � � � � { }3 2 1 0 1 2 3, , , , , ,B B B B B B B− − −� � � � � � �
{ }3 2 1 0 1 2 3, , , , , ,C C C C C C C− − −� � � � � � � and{ }1 2 3, ,D D D� � � .
Choose seven linguistic variables for ,n ns s� and nε . The
linguistic variables are defined as follows:
e
, kε ,s s�
r
δd
SMC
FC
Electric
Actuator
Fig.1. Scheme of fuzzy sliding mode controller
978-1-4244-6044-1/10/$26.00 ©2010 IEEE
{Positive Big (PB), Positive Medium (PM), Positive Small
(PS), Zero (ZE), Negative Small (NS), Negative Medium
(NM), Negative Big (NB)}.
Where, the corresponding linguistic values of
, , , ( 3, 2, 1,0,1, 2,3)i i iA B C i = − − −� �� are NB, NM, NS, ZE, PS,
PM, PB. Choose three linguistic variables of nk as {Low (L),
Medium (M), High (H)}, and the corresponding linguistic
values of ( 1,2,3)iD i =� are L, M and H.
Use triangle of membership functions, and they are shown
as below in figure 2 to 5.
μ
s
Fig.2. Membership functions of ns
μ
�
Fig.3. Membership functions of ns�
μ
nε
Fig.4. Membership functions of nε
Fig.5. Membership functions of nk
Then we can establish the fuzzy control rules which are
made up of a series of “IF -THEN” expressions. Notice that
the sliding mode can be divided into four states according to
figure 6: (I) 0 & 0s s> >� ; (II) 0 & 0s s> <� ;
(III) 0& 0s s< >� ; (IV) 0 & 0s s< <� . Choose the
parameters ε and k according to the distance between
system state and the switching surface and the tending speed
to the switching surface.
Fig.6. The fuzzification of switching function for a fuzzy sliding mode
controller
If the system is located in (I) 0 & 0s s> >� and
{ }n ns s PB= =� which means the system state is positive far
from the sliding surface and is trending to go away from the
surface in high speed. Thus, the control function should have
positive big trending speed and high control force. The other
situations can be analyzed in the same way. And the fuzzy
control rule can be shown in table 1 as below.
TABLE 1. Rule table for the /n nkε
ns�
ns
NB
NM
NS
ZE
PS
PM
PB
NB NB/H NB/H NB/H NB/M NM/L NS/L ZE/L
NM NB/H NB/H NB/M NM/M NS/L ZE/L PS/L
NS NB/H NB/H NM/H NS/M ZE/L PS/L PM/L
ZE NB/M NM/M NS/M ZE/M PS/M PM/M PB/M
PS NM/L NS/L ZE/L PS/M PM/H PB/H PB/H
PM NS/L ZE/L PS/L PM/M PB/H PB/H PB/H
PB ZE/L PS/L PM/L PB/M PB/H PB/H PB/H
The fuzzy control rule can be described in linguistic form
as follow:
lR : If ns is iA� , and ns� is jB� then nε is ijC� , and
nk is ijD� , ( , 3, 2, , 2,3)i j = − − " .
Define the control rule 1R A B C= × →� �� and
2R A B D= × →� � � , so the reasoning result can be shown as
follows.
( ) ( ) ( ) ( )1 2 1 2U A B R R A B R A B R
C D
⎡ ⎤ ⎡ ⎤= × = × ×⎣ ⎦ ⎣ ⎦
=
� � �� � �D ∪ D ∪ D
� �∪
( ) ( )
( ) ( )
C A A C B B C
D A A D B B D
⎧ ⎡ ⎤= → →⎪ ⎣ ⎦⎨ ⎡ ⎤= → →⎪ ⎣ ⎦⎩
� � � � �� �D ∩ D
� �� � � � �D ∩ D
So the membership functions can be shown as
( )
( )
BC A C
D B DA
z
z
μ ω ω ω
μ ω ω ω
= ∧ ∧⎧⎨ = ∧ ∧⎩
� � � �
� � � �
(18)
978-1-4244-6044-1/10/$26.00 ©2010 IEEE
Where ( )ω • is the maximum of the membership functions,
and z is the independent variable of the membership
functions.
Clarify the output of the fuzzy controller through
choosing center average defuzzification.
3
3
3
3
3
3
3
3
( )
( )
( )
( )
C
n
C
D
n
D
z z dz
z dz
z z dz
k
z dz
με
μ
μ
μ
−
−
−
−
⎧⎪ =⎪⎪⎨⎪⎪ =⎪⎩
∫
∫
∫
∫
�
�
�
�
(19)
IV. SIMULATION& ELECTRIC ACTUATOR SYSTEM
The transfer function of the electric actuator system can be
rewritten in form of differential equation as follows:
4( ) ( ) ( )VT t t K u tδ δ+ =�� � (20)
Define the error of the system as
i f i fe u u u K δ= − = − (21)
Thus,
1 ( )
1 ( )
1 ( )i
i
f
i
f
f
u e
u e
K
u e
K
K
δ
δ
δ
⎧⎪⎪⎪⎪⎨⎪⎪⎪⎪⎩
= −
= −
= −�� ��
� � �
��
(22)
Put equ. (20) and equ. (22) together, we have a new
equation
2 ( ) ( )e a e bu t F t+ = − +�� � (23)
Where, ( )F t is the generalized disturbance of the system,
and can be defined as
2 1( ) ( ) ( )i iF t u t a u t bf= + −� � (24)
Where
2
1
v
a
T
= , 4f
v
K K
b
T
= .
The maximum and minimum of the system disturbances
can be estimated as below
( )F t F≤
Define 1 2 1,x e x x= = � , furthermore a new state variable:
0x e=� .
Then the error state equation can be extended as
0 1
1 2
2 2 2 ( ) ( )
x x
x x
x a x bu t F t
=⎧⎪ =⎨⎪ = − − +⎩
�
�
�
(25)
The simulation results of unit step responses of the three
different controllers can be seen in figure.7.
Fig.7. Simulation result of unit step response
Moreover, the time domain characteristics can be shown in
table.2 as below.
TABLE 2. Time domain characteristics
Rise time(ms) Adjust time(ms) Overshoot
PID 19.41 31.78 0.1%
SMC 14.19 26.0 0.1%
FSMC 12.25 23.0 0.1%
It can be see through simulation that all of the three
controllers have small overshoot, but for response time, the
SMC is better than PID controller and the FSMC is better
than SMC, which verify the dynamic adjusting of the fuzzy
controller.
-0.2 0 0.2 0.4 0.6 0.8 1 1.2
-180
-160
-140
-120
-100
-80
-60
-40
-20
0
20
e
(a) Phase track of SMC
-0.2 0 0.2 0.4 0.6 0.8 1 1.2
-160
-140
-120
-100
-80
-60
-40
-20
0
20
e
de
(b) Phase track of FSMC
Fig.8. Simulation result of phase tracks
The FSMC is not only faster but also can weaken
chattering compared with SMC. As shown in figure 8, (a) is
the phase track of SMC. And the chattering amplitude is
978-1-4244-6044-1/10/$26.00 ©2010 IEEE
about 0.4. (b) is the phase track of FSMC. And the
chattering amplitude is about 0.02 which is much smaller than
0.4. It is obvious that the fuzzy control eliminates the
chattering problem of SMC.
V. CONCLUSION
In this paper a fuzzy sliding mode controller based on
trending law for electric actuator is studied. An integral
sliding mode controller is conducted to reduce the electric
actuator system error. The fuzzy control is developed to
configure the parameters of sliding mode control based on
exponential turning law in order to eliminate the chattering
problem. The fuzzy rules are constructed according to the
characters of the switching function and the sliding surface.
The simulation and analytic results verify the superiority of
the FSMC and its efficiency of weakening the chattering.
REFERENCE
[1] V. A. Utkin, “Variable structure systems with sliding mode,” IEEE Trans.
Automat. Contr., vol. AC-22, no. 2, 1977, pp. 212-222.
[2] J. Rolf, "Adaptive Controller of Robot Manipulator Motion.'' IEEE Trans.
on Robotic and Automation, Vol. 6, No. 2. 1990.
[3] Q. P. Ha, and M. Negnevitsky, “A robust modal controller with fuzzy
tuning for multimass electromechanical systems,” Proc. IEEE 3rd
Australian and New-Zealand Conf. On intelligent information systems,
Perth, Australia, 1995, pp. 412-219.
[4] W.A. Deabes, Fyez Areed. Fuzzy Sliding Motion Controller for
Six-Degree-of-Freedom Robotic Manipulator. IEEE 41st Southeastern
Symposium on System Theory University of Tennessee Space Institute
Tullahoma, TN, USA, March 15-17, 2009, pp.148-152.
[5] Xiaojiang Zhang, Zhihong Man. A' New Fuzzy Sliding Mode Control
Scheme. IEEE Proceedings of the 3d World Congress on Intelligent
Control and Automation. Hefei, P.R. China. June 28-July 2,2000,
pp.1692-1696
[6] JIANG K, ZHANG J G, CHEN Z M. A new approach for the sliding
mode control based on fuzzy reaching law[C]. Proc of the 4th World
Congress on Intelligent Control and Automation. Shanghai, China: Press
of University of Science and Technology of China, 2002, 6: 656-660.
[7] Jurgen Ackermann and Vadim I. Utkin.Sliding Mode Control Design
Based on Ackermann’s Formula[C].Lake Buena Vista, FL: 1994
[8] X.J. Liu, P. Guan, and J.Z. Liu. Fuzzy Sliding Mode Attitude Control of
Satellite. Proceedings of the 44th IEEE Conference on Decision and
Control, and the European Control Conference 2005 Seville, Spain,
December 12-15, 2005, 1970-1975.
[9] Hai-Ping Pang, Cheng-Ju Liu and Wei Zhang. Sliding Mode Fuzzy
Control with Application to Electrical Servo Drive. Proceedings of the
Sixth International Conference on Intelligent Systems Design and
Applications (ISDA'06). 2006.
[10] J. C. Wu and T. S. Liu. A Sliding-Mode Approach to Fuzzy Control
Design. IEEE TRANSACTIONS ON CONTROL SYSTEMS
TECHNOLOGY, VOL. 4, NO. 2, MARCH 1996,pp.141-151.
978-1-4244-6044-1/10/$26.00 ©2010 IEEE