You are on page 1of 4

2015 International Conference on Intelligent Transportation, Big Data & Smart City

Heavy Truck EPS System Simulation Based on PSO Optimization Fuzzy Control

Gong Jian-qiang1, Yang Wei2, Heng Wei-wei3


1
Research Institute of Highway Ministry of Transport, Beijing 100088, China;
2
Department of vehicle operation engineering Chang’an University, Xi’an,710064,China
Jilin University, 130025, Changchun, China.
E-mail: g02120103@126.com

Abstract—Based on particle swarm optimization of heavy


truck EPS system control strategy is proposed, on the basis of
vehicle EPS system overall design, taking DC motor current as
the control target, using particle swarm optimization fuzzy Ffh  F fh cos 
control strategy, using MATLAB/Simulink as simulation
platform, taking dump truck as research type models and Ffm  F fm cos 
commercial vehicle simulation model which equipped with
EPS system is established. The simulation experimental results
F
fh  Fvfh cos 
show that compared with the conventional fuzzy control, F
fm  Fvfm cos 
particle swarm optimization fuzzy control can improve the
whole dynamic response of heavy truck EPS system efficiently, DFf  F
fh  F
fm
also good steering portability and handling stability is obtained.

Keywords- heavy truck; assistance characteristics curve;


PSO optimization; fuzzy control

I. INTRODUCTION
In the field of passenger vehicles ESP system control,
research institutes and experts and scholars in China and
abroad has made a lot of research and got some results, and
mainly used PID control algorithm, fuzzy PID control
algorithm, neural network PID control algorithm, neural
network control algorithm, sliding model control algorithm
and HĞcontrol theory, etc. Some research results have been
got in the above research, but some problems still existed,
such as: whole vehicle model mostly applies 2-DOF two 
wheels model, which can’t simulate the impact of ground Fig 1. Vehicle forece analysis
resistance moment act on the steering system when moving
or pivot steering. The nonlinear state of heavy truck can’t be B. Tire model
simulated when applying linear tire model. Control Tire model is chosen as nonlinear Dugoff model
algorithm can’t be adjusted in real-time according to road Fxij  f ij C xij Sij (ij  fl , fr , rlˈrr ) (4)
and operation condition for the best control, once it has
been set[1-4]. Fyij  fij C yij ij (ij  fl , fr , rlˈrr )
II. SYSTEM OVERALL DESIGN AND DYNAMIC Frij  (Cxij Sij ) 2  (C yij ij ) 2
MODELING
 ij Fzij
A. Vehicle forece analysis
 1, F rij

  2
m v o   DF f cos   DFr  Fd fij   (5)
(1) F F F
(2  ij zij ) ij zij , Frij  ij zij
mv o   DF f sin   2 Frij 2 Frij 2
(2)
Where, C xij , C yij are the longitudinal and lateral stiffness of

mV oz  Fzfh  Fzfm  Fzrh  Fzrm  mg (3) tires; Fzij is the vertical load of tires; ij is the side-slip

978-1-5090-0464-5/16
/16 $31.00 © 2016 IEEE
$31.00 © 2016 IEEE 446
DOI 10.1109/ICITBS.2015.116
angle of tires; ij is the adhesion coefficient; S ij is the slip Fuzzy controller
+ e
ke
ratio of tires.
̶
Vertical load of each tire: kI I
ec
lr  hg  h
lr d/dt kec
Fzfl  mg  mVx  mVy 
g

2(l f  lr ) 2(l f  lr ) B f (l f  lr )
PSO
lr  hg  hg lr Fig 3. Fuzzy Control Strategy Based on Particle Swarm
Fzfr  mg  mVx  mVy 
2(l f  lr ) 2(l f  lr ) Bf (l f  lr ) Optimization
The control variable e, ec and I were quantified as [-6,
lf  hg  hg lf 6], [- 0.3, 0.3] and [-6, 6], the fuzzy subset are {NB, NM, NS,
Fzrl  mg  mVx  mVy  ZO, PS, PM, PB}. The elements of subset represent the
2(l f  lr ) 2(l f  lr ) Br (l f  lr ) negative big, negative middle, the negative small, zero,
lf  hg  hg lf positive small, positive middle, positive big. Membership
Fzrr  mg  mVx  mVy  function of input and output variables of the fuzzy
2(l f  lr ) 2(l f  lr ) Br (l f  lr ) controller applied triangular membership function.
Specified the steering wheel clockwise is positive, when e
C. DC motor model and ec is larger, if the values of both were PB, then it was in
EPS system power motor is the kind of permanent need to output larger negative assistance moment; and when
magnet brushless DC motor, with the aim to control the e and ec is smaller, if the values of both were NS, then it
terminal voltage of DC motor. DC motor simplified was in need to output smaller positive assistance moment.
equivalent structure is shown in Fig 2, The mathematical PSO implements individual and global optimization by
model of DC motor is established according to the particle population to get the best optimization, and it’s
characteristics of the circuit: suitable for nonlinear systems [7].
dI d The composition of the particle swarm given as
U L  RI  K b m (6) follows:
dt dt
x  {x1 , x2 , L , xm }
Where, K b is the back EMF coefficient;  m is the motor
The position and velocity information of each particle
rotation angle. is:
xi  {xi1 , xi 2 , L , xin }
vi  {vi1 , vi1 , L , vin }
The relationship of k+1-dimensional in PSO as
follows:
vid (k  1)  vid (k )  c1rand ()( Pid (k )  xid (k ))
(7)
 c2 rand ()( Pgd (k )  xid (k ))
xid (k  1)  xid (k )  vid (k  1) (8)
Fig 2. The Equivalent Structure of DC Motor
Pid (t )  ( Pi1 (t ), Pi 2 (t ), L , Pid (t ))
III. PARTICLE SWARM OPTIMIZATION (PSO) FUZZY
CONTROL STRATEGY Pgd (t )  ( Pg 1 (t ), Pg 2 (t ), L , Pgd (t ))
this paper got targeted control current from the assist Where,  is the inertia weight; vid (k ) is the particle
characteristic curves, by controlling the deviation between velocity of k-dimensional space; xid ( k ) is the particle
the target current and the actual current, using weighting
factors k e k ec k I of the fuzzy control strategy based on position of k-dimensional space; c1 , c2 are the acceleration
PSO online adjustment fuzzy controller to control the factors; rand () is the random number; Pid (k ) , Pgd ( k ) are
assistance current. Fuzzy control strategy based on particle respectively the individual extremum of particles and global
swarm optimization is presented in Fig 3. extremum of populations.
In order to accelerate the process of optimizing the
particle, the fitness function has been established according
to the literature [8], as follows:

447

Model parameters Specific values
f   k y(k ) dk (9) Distance from centroid to the front 1760
0 axle /mm
Where, y(k ) is the current deviation of k-moment; Tread /mm 1450
Z-axis moment of inertia /kg.m2 950
 is the weighting coefficient. Cornering stiffness of front wheel
The initialization parameter settings in PSO: the -67830
N/rad
initialization number of particles is 30, the maximum Cornering stiffness of rear wheel
-125715
number of iterations is set to 150, the value of the inertia N/rad
weight  is 0.9, the value of the particle velocity is [-3, 3], gear ratio of steering system 23
the value of particle position [0, 3], both of acceleration Gear ratio of worm reducer 20
Moment of inertia of DC motor
factors c1 , c2 are 1, the fuzzy control scale 0.009
/kg·m2
Torque coefficient of DC motor
factors k e , k ec , k I are 30 sets of random values in [0, 6]. Nm/A
0.0730
The weighted Coefficient   is 0.6. The flow chart of whole Inductance of DC motor N·m·s/rad 0.02
process of PSO is shown in Fig 4. Back-EMF coefficient V·s/rad 0.015
Serpentine test peg layout diagram as Fig 5.
Start

Initialize PSO

Set the initial position and velocity of particle Fi


g 5. tpylon couse slalom test Set
Fig 6 and Fig 7 respectively show the simulation
Calculate the particle fitness function results of yaw rate and sideslip angle response under the
state of applying fuzzy control and PSO fuzzy control, as it
adjusted according to the fitness function can be seen from the figures, compared with the
conventional fuzzy control, after applied with PSO, the self-
Update the position and velocity of particle tuning speed of yaw rate and sideslip angle response has
been faster, maximum overshoot amplitude has been
N significantly reduced, dynamic control effect has been
Meet the termination
requirements? significantly improved.
Y
Output the optimal control variables

End
Fig 4. Particle Swarm Optimization Flow Chart

IV. SIMULATION ANALYSIS


To show superiority and robustness of PSO fuzzy
control, a heavy truck EPS system simulation model has
been built in MATLAB / Simulink environment, research
models was chosen as a light dump trucks with load of
7715kg, of which the front axle load is 2750kg, then the
simulation results of conventional fuzzy control and fuzzy
control including PSO was comparatively analyzed, parts of Fig 6. The Yaw Rate Response Simulation Result
the simulation parameters of heavy truck EPS are shown in
Table 1.
Table 1. parts of simulation parameters of heavy truck EPS system
Model parameters Specific values
load of vehicle /kg 7715
Wheelbase /mm 2130
Distance from centroid to the front 370
axle /mm

448
REFERENCE
[1] Yu Jian-cheng. The Study Based on PID Control Strategy of
the Automotive Electric Power Steering System [J]. Journal
of Hebei Institute of Technology, 2005 (1): 63-65.
[2] Yang Wei. Research on Fuzzy Control Strategies of
Commercial Vehicle EPS System [D]. Xi’an: Chang’an
University, 2010.
[3] Tian Fengnian, Lu Zhixiong, Zhao Lanying, et al. A New
Bectro-controlled Hydraulic Power Steering System Based
on PIDNN Algorithm [J]. Journal of Highway and
Transportation Research and Development, 2010, 27(8):
136-142.
[4] Wang Huaqiang, Yuan Hao, Yang Tuguang. Application of
Adaptive Fuzzy Neural Network in EPS [J]. Journal of Hefei
University of Technology: Natural Science, 2011, 34(2):
188-191.
Fig 7. The Vehicle Slip Angle Response Simulation Result [5] Zhao Zhiguo, Yu Zhuoping, Sun Zechang, et al. A Research
Fig 8 shows the simulation results of vehicle path on HĞ Robust Control Strategy for Electric Power Steering
following response at the conditions before and after added System [J]. Automotive Engineering, 2006, 27(6): 730-735.
PSO fuzzy control, as shown in the figure, after applied [6] Zhang Shuo, Yu Qiang, Yan Guanghui, et al. A Research on
the Power-assistance Control Strategy for the EPS in a
with PSO control, the lateral offset between the vehicle path
Commercial Vehicle [J]. Automobile Engineering, 2013 (9):
and the actual path has been smaller, vertical offset is within 812-816.
the allowable range, the able to follow designed line is [7] Kim J H, Song J B. Control Logic for An Electric Power
better as well, and the entire vehicle obstacle avoidance Steering System Using Assist Motor [J]. Mechatronics, 2002,
capability has been effectively improved, the handling 12(3): 447-459.
stability and driving safety of the vehicle has been improved [8] Sun Lijun, Chen Long, Zhao Jingbo, et al. The Electric
significantly. Power Steering System Dynamical Characteristics Research
and Test [J]. Journal of Zhengzhou University: Engineering
Science, 2007, 28(4): 117-121.
[9] Sun Bo, Chen Weidong, Xi Yugeng. Particle Swarm
Optimization Based Global Path Planning for Mobile Robots
[J]. Control and Decision, 2005, 20(9): 1052-1060.
[10] Meng Qingkuan, Qiu Ruicheng, Zhang Man, et al.
Navigation System of Agricultural Vehicle Based on Fuzzy
Logic Controller with Improved Particle Swarm
Optimization Algorithm [J]. Transactions of the Chinese
Society of Agricultural Machinery, 2015, 46(3): 29-36.

Fig8. The Path Following Response Simulation Result

V. CONCLUSIONS

This paper estabished the vehicle fore modelǃ tire


model and the DC motor model. Assistance characteristic
curve has been developed based on experimental heavy
truck steering system test. Taking the DC current as the
control target, a PSO fuzzy control strategy has been
developed. A heavy truck EPS system simulation model has
been established. The results show that, PSO fuzzy control
was able to achieve better dynamic control effect, and the
steering agility, handling stability and driving safety of
vehicles have been improved about 10%.

449

You might also like