Indonesian Journal of Electrical Engineering and Computer Science
Vol. 14, No. 2, May 2019, pp. 903~911
ISSN: 2502-4752, DOI: 10.11591/ijeecs.v14.i2.pp903-911
903
A simplify fuzzy logic controller design based safe
experimentation dynamics for Pantograph-Cateary system
Mohd Faisal Farhan, Nor Sakinah Abdul Shukor, Mohd Ashraf Ahmad, Mohd Helmi Suid,
Mohd Riduwan Ghazali, Mohd Falfazli Mat Jusof
Faculty of Electrical and Electronics Engineering, Universitas Malaysia Pahang, Malaysia
Article Info
ABSTRACT
Article history:
Contact force between catenary and pantograph of high speed train is a
crucial system to deliver power to the train. The inconsistence force between
them can cause the contact wire oscillate a lot and it can damage the
mechanical structure of system and produce electric arc that can reduce the
performance of system. This project proposes a single-input fuzzy logic
controller (SIFLC) to control the contact force between the pantographcatenary by implement Safe Experimentation Dynamics (SED) method to
tune the SIFLC parameters. The essential feature of SIFLC is that it is
model-free type controller design with less pre-defined variables as
compared to other existing model-based controllers. The performance of the
SIFLC is analyzed in terms of input tracking of contact force of pantographcatenary and time response specifications. A simplified model of three
degree of freedom (3-DOF) pantograph-catenary system is considered.
In this study, the simulation result shows that the SIFLC successfully track
the given contact force with less overshoot with percentage different of peak
to peak response from actual force 2% and fast response within 5.27s.
Received Sep 25, 2018
Revised Nov 26, 2018
Accepted Dec 6, 2018
Keywords:
Contact force
Fuzzy logic control
Pantograph-catenary
Safe experimentation dynamic
Copyright © 2019 Institute of Advanced Engineering and Science.
All rights reserved.
Corresponding Author:
Mohd Faisal Farhan bin Kamerul Bieza,
Faculty of Electrical and Electronics Engineering,
Universiti Malaysia Pahang,
26600 Pekan, Pahang, Malaysia.
Email: mohdfaisalfarhan@ump.edu.my
1.
INTRODUCTION
Public transportation is one of the crucial systems for the community to ease their life day by day.
Along the advancement of technologies, some electronic device is use to control the system of the public
transport such as high speed train. This electronic device used to optimum the train performance and services.
In high speed train system, the functions of electronic control device are to control the contact force between
catenary and pantograph. Pantograph-catenary (PAC) system will determine the amount of electricity
required to power the motors of an electric train.
The main problem on high-speed train pantograph-catenary system is to control the contact force
between them. When there is a sliding movement between the pantograph and catenary, a periodic vibration
of the pantograph will form an inconsistence in the contact force between the PAC systems due to the change
of stiffness. This inconsistence force will generate an electric arc that can damage the mechanical structure of
the PAC systems and reduce the system performance.
In most of the cases, the desired constant contact force that is required by the PAC system in all
operating condition is around 100N [1]. However, in order to get the actual value for the constant contact
force, the counteraction of the variation contact force could be obtained by using several methods. One of the
methods is by tightening the contact wire tension to increase the stiffness. Besides, the other method is by
increasing modulus of the contact force [2]. The first method actually will be more costly because the railway
Journal homepage: http://iaescore.com/journals/index.php/ijeecs
904
ISSN: 2502-4752
systems need to be connected at the same standard. By considering this situation, other method was
implemented which by using a controller to obtain the constant contact force.
So far, there are many controllers have been reported to control the contact force between the
catenary and pantograph. There are active control [3], adaptive fuzzy logic control [4] and fuzzy sliding
mode controller [5]. Most of the developed controllers are model-based controller that requires an accurate
model of the system, while there is always a gap between the model and the real system, such as unmodeled
dynamics problem. Moreover, the developed conventional fuzzy logic in [4] requires a lot of effort to find the
optimal variables, such as types of membership functions (MFs), range of universal boundary, number of MF
and many more. On the other hand, the SIFLC is a good potential on solving above mentioned issues.
The SIFLC does not require any knowledge of the plant in its design procedure and it requires less predefined variables than the conventional FLC. In particular, it does not require any rules and membership
function. So far, the SIFLC has been implemented in various type of control problems, such as flexible joint
manipulator of robot [6], double pendulum type overhead crane [7], underwater vehicle manipulator
system [8] and magnetic levitation system [9].
In this paper, a single input fuzzy logic controller (SIFLC) of pantograph-catenary for high-speed
train is proposed. The simplification of conventional FLC is called as a single input fuzzy logic (SIFL).
The simplification of SIFLC is formed by using a signed distance approach [10]. The reduction in the number
of rules in rule table to one-dimension allowing it to be treated as single input single output (SISO)
controller. The tuned parameter for SILFC is determined by using SED method [11]. The proposed SIFLC is
tested to a 3-DOF PAC system. The performance of SIFLC is analyzed in terms of trajectory tracking of
contact force and its time response specifications.
2.
MODELLING PAC SYSTEM
The pantograph-catenary system model considered in this work is shown in Figure 1. The mass at
the top represent by
, the mass in the middle represent by
and the lower mass represent by
.
In the normal condition, the catenary will be in contact with the pantograph. In this condition
and the
weight of upper block will be
where the weight of catenary is
. Therefore, the state-space
equations are shown in (1) and (2).
x Ax Bu Dw(t )
(1)
y Cx,
(2)
̇
with the vector
̇
̇
, and matrices , ,
and
are expressed by,
1
0
0
0
0
0
k2
b2
0
0
22
21
m1
m1
0
1
1
0
0
0
k
b
b
b
A(t ) 1
1
1
1
43 44
m2
m2
m2
m2
0
0
0
0
0
1
k2
b2
65 66
0
0
m3
m3
where:
21
k1 k c
m1 mc
,
b1 bc ,
m1 mc
k k2 ,
1
m2
22
43
44
b1 b2
m2
k3 k 2
m3
b3 b2
m3
65
,
66
,
,
Indonesian J Elec Eng & Comp Sci, Vol. 14, No. 2, May 2019 : 903 – 911
(3)
Indonesian J Elec Eng & Comp Sci
B 0
0
C Q1
D 0
0
Q2
0
0
0
0
Q3
0
1
m3
Q4
0
1
m3
ISSN: 2502-4752
905
T
0
(4)
0
T
(5)
T
(6)
where,
m1 k c (2 m1 mc ) k1
m1 mc
m1 bc (2 m1 mc ) b1
Q2
m1 mc
(2 m1 mc ) k1
Q3 Q4
m1 mc
Q1
To get more accurate result on modeling the PAC system, mass and damper variation are
considered. The catenary is modeled by the system spring, damper and mass. The mass
, spring stiffness
and damping constant , exhibit periodic characteristic for each section. Therefore, a Fourier series
expansion is considered as shown in (7).
2i
3
mc (t ) mc 0 i 1 mci cos( x(t ))
L
2i
3
bc (t ) bc 0 i 1 bci cos( x(t ))
L
2i
3
kc (t ) kc 0 i 1 kci cos( x(t ))
L
(7)
where the tower and the train is separated with actual distance of
,the span length represent by, .
The train speed is represented,
where the velocity of train is, and is time to reach the
distance. For each of the parameter in PAC system, we just estimate it by referring to the previous research
paper in [5]. The baseline catenary parameters that have been used in the pantograph and catenary model is
shown in Table 1.
Table 1. Baseline Catenary Parameters [11]
Parameter
Notation
Value
Catenary Mass
Catenary Dampers
Catenary Stiffness
Span Length
Figure 1. PAC system linear model
A simplify fuzzy logic controller design based safe experimentation dynamics for… (Mohd Faisal Farhan)
906
ISSN: 2502-4752
3.
SINGLE INPUT FUZZY LOGIC CONTROLLER (SIFLC)
This section presents the design of SIFLC. Most of the FLC consist of two variable inputs,
which are error ( ) and error different ̇ ). Instead of using two input variables ( ̇ ) to produce the output
, it is possible to use a single input variable only. The application of SIFLC will reduce the number of rules
compared to the case of conventional Fuzzy Logic controller.
The single input fuzzy logic can be obtained by using a method called signed distance approach.
This approach will remove the number of input into single input variable called signed distance. The distance
to an actual diagonal line is called signed distance where the input variable ( ̇ ), come out from diagonal
line, . In other words, to derive the distance variable, Q( ̇ ) need to be an intersection point at
and
perpendicular to operating point P( ̇ ) [6].
From the Figure 2, the straight line function can be represented by . The equation of the straight
line is:
Figure 2. Signed distance approach
̇
(8)
From (8), is the inclination magnitude of the diagonal line
defined as [6].
d
. The distance,
from P(
e e
̇ ) to Q(
̇ ) can be
1 2
(9)
The result from the derivation of input distance variable will form a one-dimension rule table instead
of two-dimension rule table that was used in conventional FLC [10]. The rule table before the input variable
derivation process is shown in Table 2 while the simplified rule table is depicted in Table 3 where
and
are the diagonal. The diagonal line related to the new input for rule table, while
and
represent the output related to diagonal line. The value of determines the
control action of FLC. Therefore, it is appropriate to call this controller as SIFLC.
Table 2. Two-dimension Rules Table
̇
LN
MN
SN
ZE
SP
MP
LP
LP
ZE
SP
MP
LP
LP
LP
LP
MP
SN
ZE
SP
MP
LP
LP
LP
SP
MN
SN
ZE
SP
MP
LP
LP
ZE
LN
MN
SN
ZE
SP
MP
LP
SN
LN
LN
MN
SN
ZE
SP
MP
MN
LN
LN
LN
MN
SN
ZE
SP
Table 3. One-dimension Rules Table
LN
LN
LN
LN
LN
MN
SN
ZE
μo
LLN
LN
LMN
MN
LSN
SN
LZE
ZE
LLP
SP
LMP
MP
LLP
LP
Figure 3 shows the structure of control system for pantograph-catenary system in high speed train
using SIFLC. Figure 4 shows the detailed structure of SIFLC. It consists of several blocks that is derived
from (9). The control surface is designed based on a saturation function that can be amplified by using a
gain r.
Indonesian J Elec Eng & Comp Sci, Vol. 14, No. 2, May 2019 : 903 – 911
Indonesian J Elec Eng & Comp Sci
ISSN: 2502-4752
907
Figure 3. Structure of control system
Figure 4. SIFLC structure
4.
A.
SIFLC DESIGN BASED SAFE EXPEIMENTATION DYNAMICS
A data-driven SIFLC tuning based on SED algorithm is stated in this section.
Safe Experimentation Dynamics (Sed)
The problem of the optimization is stated as
(10)
where the function represents by
and
is parameter design. From (10), the optimal solution
is achieved by updating the parameter design using SED algorithm. The rules in updating SED
algorithm is stated as
̅
(11)
where
is iteration number,
is the th element of
, ̅
is the th iteration of
̅
. ̅ used to save current best value of design parameter, is the size of interval to determine the
random steps on
and the random number value is
. In (11), the function is given by
{̅
̅
̅
̅
(12)
where the pre-determined minimum and maximum parameter design is and
and
. The steps
procedures to implement SED algorithm are as follows:
Step 1: Define the value of
, and . is set
and
set as the initial condition for
parameter design. The loss function be
) and as normal ̅
and ̅
. For a new random
setting for , is probability that used where it is scalar. To save the loss function present best value,
variable ̅ is used.
Step 2: When the value
̅ , execute ̅
and ̅
. Apart of that, proceed to
next step.
Step 3: Produce random number . Produce second random number if
and by updating the (11),
obtained the value of
. Otherwise,
=̅ .
)
Step 4: Achieve the loss function (
Step 5: If the pre-stated termination condition that based on the designated maximum number of iteration,
is satisfied, the algorithm stops with the solution
. Otherwise,
{
}
repeat step 2.
A simplify fuzzy logic controller design based safe experimentation dynamics for… (Mohd Faisal Farhan)
908
B.
ISSN: 2502-4752
Siflc-Tuned Sed
For this section, the SIFLC-tuned SED design parameter is presented by
(13)
. Then the
In particular, we set
and the loss function is
∫
SIFLC-tuned SED can be implemented as follows:
Step 1: Consider
and
. Then, the maximum iteration,
is determine.
Step 2: Execute the SED algorithm in Section 4.1.
Step 3: When reaching
, attain the optimal output
̅
. Then implement
the SIFLC system in Figure 4.
in
5.
SIMULATION RESULTS
In this section, the simulation result of the SIFLC for controlling the contact force between the
catenary and pantograph is presented. The SED algorithm parameters are set as
,
,
,
and
. The train velocity is set as
and the desired input
function is given by
(14)
The loss function response when 150 iterations have been used for SED is shown in Figure 5.
The loss function is successfully minimized after 150 iterations. Table 4 shows the optimal design
parameters. From the results, it exposed that SED method can minimize the loss function.
Table 4. Design Parameter
correspond to
SIFLC Gain
r
-2
0
0.01
1
correspond to
-1.5318
0.6330
0.0294
4.2958
Figure 5. Response of loss function using SED
Next, the response of SIFLC is shown in Figure 6. The blue line represents desired response for the
PAC system while the red lines indicate the response after 150 iteration of SED. It shows that the response
can track the desired reference signal. There are some oscillations after 3.6 seconds and this result is
comparable with the result in [5].
Indonesian J Elec Eng & Comp Sci, Vol. 14, No. 2, May 2019 : 903 – 911
Indonesian J Elec Eng & Comp Sci
ISSN: 2502-4752
909
Figure 6. SIFLC performance when tuned by SED
In Figure 7, progress of denote that permanents contact of the PAC. The observation oscillation
shows the vertical movement of pantograph floors. This oscillations influence by parameters that depends on
the displacement of the train. Therefore, they simulate the fluctuations that can be determined in actual
simulation of PAC.
Figure 7. Displacement of the masses of the model
From Figure 8, the performance of SIFLC is compared to PID controller. For SIFLC the response
shows it start to response smoothly until reach the desired contact force which is 100N. On the other hand,
the oscillation of the tracking contact force is almost same as compared to SIFLC. As resulted, SIFLC have
the fast response as compared to PID controller.
Figure 8. Output tracking for SIFLC and PID controller
A simplify fuzzy logic controller design based safe experimentation dynamics for… (Mohd Faisal Farhan)
910
ISSN: 2502-4752
Table 5 shows the controller performance for SIFLC and PID. As shown in the table, SIFLC have a
fast response as compared to PID controller because of the rise time to track the desired contact force.
However, both controllers have the same percentage overshoot due to the same oscillation on tracking
desired contact force.
Table 5. Controller Performance
Parameter
Rise Time
Percentage Overshoot
Peak to Peak Value of Force at Steady State
Percentage Different Between Peak to Peak
SIFLC
5.27s
1%
98N to 102N
2%
PID
6.22s
1%
95N to 102N
3%
6.
CONCLUSION
From this study, the problem on the inconsistence contact force between the catenary and
pantograph is exposed and a solution based on SIFLC is presented and compared with the PID controller
performance. The SED algorithm can improve the tuned parameters that need to be used by SIFLC.
The simulation is applied to the time variable, consideration of the uncertain pantograph-catenary and a speed
of 350
.
ACKNOWLEDGEMENTS
This project has been fully supported by UMP research grant PGRS1703104.
REFERENCES
A. Levant, A. Pisano, and E. Usai, “Output-feedback control of the contact-force in high-speed-train pantographs,”
in Proceedings of the 40th IEEE Conference on Decision and Control (Cat. No.01CH37228), 2001, vol. 2,
pp. 1831–1836 vol.2.
[2] G. Poetsch et al., “Pantograph/catenary dynamics and control,” Veh. Syst. Dyn., vol. 28, no. 2–3,
pp. 159–195, 1997.
[3] W. Shudong, G. Jingbo, and G. Guosheng, “Research of the active control for high-speed train pantograph,” in
2008 IEEE Conference on Cybernetics and Intelligent Systems, 2008, pp. 749–753.
[4] E. Karaköse and M. T. Gençoğlu, “Adaptive fuzzy control approach for dynamic pantograph-catenary interaction,”
in Proceedings of 15th International Conference MECHATRONIKA, 2012, pp. 1–5.
[5] N. Mokrani and A. Rachid, “A robust control of contact force of pantograph-catenary for the high-speed train,”
in 2013 European Control Conference (ECC), 2013, pp. 4568–4573.
[6] G. Mohd Riduwan, S. Mohd Helmi, I. Zuwairie, S. Muhammad Salihin, and M. T. Mohd Zaidi, “Single Input
Fuzzy Logic Controller for Flexible Joint Manipulator,” Int. J. Innov. Comput. Inf. Control, vol. 12, no. 1, pp. 181–
191, Feb. 2016.
[7] M. A. Ahmad, M. S. Saealal, R. M. T. R. Ismail, M. A. Zawawi, A. N. K. Nasir, and M. S. Ramli, “Single input
fuzzy controller with command shaping schemes for double‐pendulum‐type overhead crane,” AIP Conf. Proc., vol.
1337, no. 1, pp. 113–117, Jun. 2011.
[8] P. S. Londhe, M. Santhakumar, B. M. Patre, and L. M. Waghmare, “Task Space Control of an Autonomous
Underwater Vehicle Manipulator System by Robust Single-Input Fuzzy Logic Control Scheme,” IEEE J. Ocean.
Eng., vol. 42, no. 1, pp. 13–28, Jan. 2017.
[9] K. Ishaque, Y. Saleem, S. S. Abdullah, M. Amjad, M. Rashid, and S. Kazi, “Single input fuzzy logic controller for
magnetic levitation system,” in Simulation and Applied Optimization 2011 Fourth International Conference on
Modeling, 2011, pp. 1–6.
[10] B.-J. Choi, S.-W. Kwak, and B. K. Kim, “Design and stability analysis of single-input fuzzy logic controller,” IEEE
Trans. Syst. Man Cybern. Part B Cybern., vol. 30, no. 2, pp. 303–309, Apr. 2000.
[11] N. S. A. Shukor, M. A. Ahmad, and M. Z. M. Tumari, “Data-driven PID tuning based on safe experimentation
dynamics for control of liquid slosh,” in 2017 IEEE 8th Control and System Graduate Research Colloquium
(ICSGRC), 2017, pp. 62–66.
[1]
Indonesian J Elec Eng & Comp Sci, Vol. 14, No. 2, May 2019 : 903 – 911
Indonesian J Elec Eng & Comp Sci
ISSN: 2502-4752
911
BIOGRAPHIES OF AUTHORS
Mohd Faisal Farhan Kamerul Bieza received his B.Eng (Hons.) in Electrical and Electronics
Engineering from Universiti Malaysia Pahang (UMP) in 2018. He is currently, an Electrical
Engineer at NDI Automation Pte. Ltd, Singapore and involve in automation industry.
Nor Sakinah Abdul Shukor received his first degree in B.Eng Electrical (Electronics) from
University Malaysia Pahang (UMP) in 2012. Now, she further her study in Master of Science at
Univerity Malaysia Pahang (UMP). Her current research field are control underactuated systems,
and game theoric methods.
Mohd Ashraf Ahmad received his first degree in B.Eng Electrical Mechatronics and Master
Degree in M.Eng Mechatronics and Automatic Control from University of Technology Malaysia
(UTM) in 2006 and 2008, respectively. In 2015, he received a Ph.D in Informatics (Systems
Science) from Kyoto University. Currently, he is a lecturer in Faculty of Electrical and
Electronics Engineering, University Malaysia Pahang (UMP). His current research interests are
model-free control, control of mechatronic systems, nonlinear system identification and vibration
control.
Mohd Helmi Suid received the B.Eng (Hons.) in Electrical and Electronics Engineering from
Universiti Teknikal Melaka in 2007. In 2014, he receive M.Sc from School of Electronic and
Electrical Engineering, University Sains Malaysia, Penang. He is currently, a lecturer in Faculty
of Electrical and Electronics Engineering, University Malaysia Pahang (UMP). His current
research interests are digital image processing, vision for robotics application and computational
intelligence.
Mohd Riduwan Ghazali received B-Eng in Electical Mechatronics August 2007 and M-Eng in
(Mechatronics & Automatic Control) at UTM Malaysia Mei 2010. Currently, he was lecturer
Faculty of Electrical and Electronics Engineering, University Malaysia Pahang (UMP).
His research interest in data-driven control, mechatronics system, robotics and control
instrumentation.
Mohd Falfazli bin Mat Jusof is a Lecturer in Faculty of Electrical and Electronics Engineering,
University Malaysia Pahang (UMP). He was graduated from Ehime University, Japan for
Bachelor (2008) and Master Degree (2010). His current research interests are IoT, application on
embedded system, computer vision system and physical computing.
A simplify fuzzy logic controller design based safe experimentation dynamics for… (Mohd Faisal Farhan)