You are on page 1of 5

TECHNICAL NOTE 1265

Model-based development for an


electric power steering system
W Ren1,2∗ , H Chen1,2 , and J Song2
1
CAD Center, Huazhong University of Science and Technology, Wuhan, Hubei, People’s Republic of China
2
State Key Laboratory of Automotive Safety and Energy Conservation, Tsinghua University, Beijing,
People’s Republic of China

The manuscript was received on 27 October 2007 and was accepted after revision for publication on 25 February 2008.
DOI: 10.1243/09544062JMES925

Abstract: A model-based development method for electric power steering (EPS) system has
been explored. A practicable model for the EPS system has been established in a full vehicle
mechanical system environment. The performance of the electric control system of the EPS
system has been evaluated in this static analysis environment. The model has then been used
in a dynamic test environment based on dSPACE hardware and software, including Software-in-
the-Loop and Hardware-in-the-Loop. The test result validates the simulation model, and shows
that this development method can be used to evaluate the conceptual design of the EPS system
as well as the control software design and testing.

Keywords: electric power steering, model-based development, dSPACE

1 INTRODUCTION developing different subsystems is also needed. There-


fore, a model-based development is a rational way for
Electric power steering (EPS) system is a very impor- this parallel developing and cooperation.
tant component for improving automotive handling In this paper, a model-based development method
and stability [1]. One of the most important parts of for EPS system has been explored. A practicable model
the EPS system is the electric control system. It can for the EPS system has been established in a full vehicle
receive the signals collected by the sensors for vehi- mechanical system environment. The electric con-
cle speed, steering angle, steering torque, etc. and trol system performance of the EPS system has been
control the assistant motor for giving the required evaluated in this environment. This model has been
assistant torque. It has been used in more and more correlated with test data using Hardware-in-the-Loop
vehicles, and can improve vehicle performance as (HiL) method and dSPACE hardware and software.
well as its efficiency and energy conservation. There
are some researches for the EPS control strategies to
improve its performance, which is crucial for the EPS 2 MODEL-BASED DEVELOPMENT PROCESS
development [2–5].
An EPS system includes the mechanical subsys- The model-based development method has been
tem and the electronic and control subsystem, and widely used for the vehicle component development,
it has to work in the full vehicle mechanical system especially for the electric control systems. The pro-
[6–8]. In the development process of the EPS sys- cesses include the development requirement from the
tem, the different subsystems have to be developed full vehicle level to the component level. A simulation
in parallel so as to reduce the time and cost of sys- model is established, and the static analysis as well
tem development. Cooperation between engineers for as the Model-in-the-Loop (MiL) simulation has been
finished for the review. Thus, the software and hard-
ware have been developed using the Rapid Control
∗ Corresponding author: CAD center, Huazhong University of Prototype tools. And the Software-in-the-Loop (SiL)
Science and Technology, 1037 Luoyu Road, Wuhan, Hubei 430074, simulation and HiL simulation have been finished in
People’s Republic of China. email: renweiqun@tsinghua.org.cn the dynamic test. All these processes, which are shown

JMES925 © IMechE 2008 Proc. IMechE Vol. 222 Part C: J. Mechanical Engineering Science
1266 W Ren, H Chen, and J Song

Fig. 2 Full vehicle’s mechanical system model

Fig. 1 Model-based development method [9] module. This method is used in vehicle dynamics
modelling, and the full vehicle model is shown in Fig. 2.
in Fig. 1, are based on the reuse of the simulation This mechanical subsystem model for the full vehi-
model [9]. cle and the steering system can be used for gathering
The method above is used for EPS development, the necessary signals from the sensors, which is the
simulation, and evaluation. The first phase of this pro- input for the control system. There are some variables
cess is the building of a reusable system-level model. setting in the steering column, shown in Fig. 2, which
The EPS model should be involved in a full vehi- perform the functions of sensors. They can collect the
cle model. The second phase is using this simulation useful data from sensor signal, which can be used in
model for MiL evaluation. The model is reviewed and the control strategy, such as the steering angle, steer-
optimized in this static analysis phase. The third phase ing torque, and vehicle speed. The mechanical model
is to transfer the optimal model to the SiL and HiL is also the final implementation for the control sys-
environment, and the performances for the control tem, which receives the control system output signal
system software and hardware can be checked and and performs correspondingly.
made sure in this dynamic test phase. After several The control subsystem model includes signals from
phases’ evaluation and optimization, the control sys- the sensors, signals to the actuators, and the control
tem software and hardware can fulfill the requirement strategies for the signals. The controller receives input
of the EPS system completely. The mechanical system signals from the sensors, and sends output signals
and control system can be jointed, and this model- to the actuators after calculating the control strat-
based development process can be used for the EPS egy. As the sensor signals are from the mechanical
system development. subsystem and the actuator signals are sent to the
mechanical subsystem, the mechanical model is cou-
pled with the controller model. The control subsystem
3 MODELLING AND STATIC ANALYSIS is modelled in MATLAB/Simulink, and the coupling
of these two sets of software is implemented by
An EPS system includes subsystems of different types. the MSC.ADAMS/Control module. The control system
Hence, the EPS system model should include these dif- model is shown in Fig. 3.
ferent types of subsystems, and it should be a hybrid In the control scheme, the shadowed square is the
model. The model should be in system level, so that block exported from MSC.ADAMS/Control. From this
it can be used for evaluating the effect of different mechanical subsystem model block, the control sys-
components for the full system. tem input, including steering angle, steering torque,
The system-level model includes a mechanical sub- vehicle speed, etc, can be collected from the sensor
system model for a full vehicle, and for the steering signals. These signals are sent to a typical proportional
system in detail. This mechanical model for the full integral derivative (PID) controller, which is widely
vehicle and the steering system is established in the used in EPS controllers. The PID controller processes
software MSC.ADAMS. It includes the suspensions for the signals from the sensors and gets a target current
the four corners. The steering system’s mechanical for the assistant motor. The target current is then com-
model includes the steering wheel, steering column, pared with the current sensor signals of the motor,
steering rack/pinion, and some connection poles. The and the difference is used by the control strategy to
chassis model is coupled with the road surface by tire produce the controller output, which is serially sent
model, which is built based on the MSC.ADAMS/TIRE to the pulse width modulation (PWM) and the motor.

Proc. IMechE Vol. 222 Part C: J. Mechanical Engineering Science JMES925 © IMechE 2008
Model-based development for an electric power steering system 1267

Fig. 3 Control system model for EPS, including full vehicle model exported from ADAMS

PWM is used to control the motor armature voltage, close-loop virtual tests, such as double lane change
which can be expressed as a lag block following the PID test, slalom test, etc. The result includes the steering
controller block in MATLAB/Simulink. The motor can torques performed by the EPS system and the hand
produce the assistant torque, which can be expressed torques implemented by the driver. The virtual test
as another lag block following the PWM block in MAT- result with EPS system is compared with the test with-
LAB/Simulink. The assistant torque produced by the out EPS system. From the result, it is obvious that the
motor performs on the steering column directly, as the steering torque from the driver’s hands can be reduced
motor is installed on the steering column. In the simu- with the EPS system assistance.
lation model above, the controller output torque signal This off-line simulation above can be described as
is sent to the mechanical system, and it performs a an MiL environment. The MiL phase can be used for
torque on the steering column model. the controller research and development. For instance,
As shown in Fig. 3, the use of MSC.ADAMS the parameters of the PID controller (Kp, Ki, Kd) can
and Matlab together results in co-simulation. How-
ever, the co-simulation is off-line simulation, as
MSC.ADAMS simulation is not a real time one. The
MSC.ADAMS/Control exported block can be inte-
grated into the MATLAB/Simulink model, and the
MATLAB/Simulink solver can handle the communica-
tion timestep issue. The mechanical system produced
differential-algebra equations together with the con-
trol system produced equations can be solved by
MATLAB/Simulink arithmetic. As the STIFF problem
can be found in some issues, an appropriate arithmetic
should be chosen for solving this problem.
This off-line co-simulation model can be used to
evaluate vehicle performance under the EPS control
system and control strategy effect. Several test condi-
tions are the same as those in the physical prototype
experiments chosen for the open-loop virtual tests,
such as the step angular input test, impulse angu- Fig. 4 The structure of the EPS test system based on
lar input test, sinusoidal angular input test, and the dSPACE

JMES925 © IMechE 2008 Proc. IMechE Vol. 222 Part C: J. Mechanical Engineering Science
1268 W Ren, H Chen, and J Song

be changed and the controller effect can be accessed real-time performance, robustness, etc. The dynamic
in the full vehicle level environment. The most fitful test begins with the download of control software code,
parameter can thus be chosen out, and the optimal which can be using the Rapid Control Prototyping
design for the control strategy can ultimately be found. tools, dSPACE. The software code downloaded by the
dSPACE system can be checked in the SiL environ-
ment, as well as the controller hardware being checked
4 DYNAMIC TEST in the HiL environment.
In the laboratory, a dynamic test environment is
The MiL analysis and reviews can get the resultant established. The dSPACE system has been used for
optimal design for the control strategy. But the result building the SiL and HiL environment. The structure
above is a theoretical one and needs to be checked of the test system based on dSPACE is shown as Fig. 4.
in the dynamic test for the performance, such as the And the test system built up according to this structure
is shown in Fig. 5.
In this dynamic test environment, an original steer-
ing wheel system is connected with steering column
and steering rack-pinion, which is connected with
two springs giving the resistant force similar to the
road load condition. This mechanical subsystem is
connected with the control subsystem, which is estab-
lished surrounding the dSPACE kernel. A duplicated
steering wheel is installed above the original steer-
ing wheel. Inside the duplicated steering wheel, there
are sensors, which can collect the control input sig-
nals such as steering angle and torque. As the test
rig is fixed on the ground, the vehicle speed imita-
tional signal has to be used from a signal producer. All
Fig. 5 Test system for EPS the input signals are sent to the dSPACE I/O channel,

Fig. 6 The test result for the assistant torque in different speeds (a) v = 0 km/h, (b) v = 10 km/h,
(c) v = 40 km/h, and (d) v = 80 km/h

Proc. IMechE Vol. 222 Part C: J. Mechanical Engineering Science JMES925 © IMechE 2008
Model-based development for an electric power steering system 1269

and the dSPACE is connected with a host computer. ACKNOWLEDGEMENTS


The MATLAB/Simulink control strategy can be down-
loaded into the dSPACE ECU, and the output actuator This paper is sponsored by the Chinese National
signal is sent to the motor. The motor is installed in Science Funding (60674067), and the opening founda-
the steering column, and it is connected to the PWM tion of State Key Laboratory of Automotive Safety and
in its frontage. The motor produced assistant torque is Energy Conservation (KF2006-02) in Tsinghua Univer-
used for reducing the hand produced steering torque, sity, Beijing, China, which is greatly appreciated. The
which is the basic function for EPS system. support from Henglong group, one of the most famous
This dynamic test environment can be used for eval- Chinese automotive component manufacturers, is
uating the control strategy and the performance of the also greatly appreciated.
EPS system. For example, the steering efforts test has
been implemented in this test system. The test con-
dition is under four different vehicle speeds, as 0, 10,
40, and 80 km/h. The steering wheel is given a sinu- REFERENCES
soidal angular input, and the sinusoidal input begins
in the 5 s point and ends in the 25 s point, which shows 1 Nakayama, T. and Suda, E. The present and future
a 20 s time period. The sinusoidal input amplitude is of electric power steering. Int. J. Veh. Des., 1994, 15,
±360◦ . The test result is shown in Fig. 6. In this result, 243–254.
Figs 6(a) to (d) show different vehicle speeds: 0, 10, 2 Badawy, A., Zuraski, J., Bolourchi, F., and Chandy, A.
40, and 80 km/h. In the figures, A is the current for the Modeling and analysis of an electric power steering
motor (Ampere), which can show the steering assistant system. SAE 1999-01-0399, 1999.
torque. B is the actual steering torque (Nm). 3 Kim, H. J. and Song, J. B. Control logic for an electric
power system using assist motor. Mechatronics, 2002, 12,
From Figs 6(a) to (d), it can be seen that the motor
447–459.
current for the steering assistant torque at 0 and 4 Yun, S. C., Cho, N. G., and Han, C. S. A simulation
10 km/h is ≈15 A, and the assistant current is reduced study of the control algorithm of a ball screw type
to 7 A at higher speed. It shows that the control strat- motor based power steering system. Proc. Instn Mech.
egy shown in Fig. 3 can be implemented in the test Engrs, Part D: J. Automobile Engineering, 2002, 216,
system shown in Fig. 5. The assistant torque can ful- 11–22.
fill the requirement in different vehicle speeds. These 5 Cabanas, R. C. and Wangb, L. Y. Control of elec-
results also show that the model-based development trical power assist systems: H ∝ design, torque esti-
method is valid in the EPS system development. mation and structural stability. JSAE Rev., 2001, 22,
435–444.
6 Haug, E. J. Computer aided kinematics and dynamics of
5 CONCLUSION mechanical systems. In Basic methods, vol. I, 1989 (Allyn
and Bacon, Boston).
1. The model-based development method for the EPS 7 Gillespie, T. D. Fundamentals of vehicle dynamics. SAE
system has been explored and can be validated. No R-114, 1992.
8 Kristine, M., Tanaka, H., and Inoue, N. An EPS control
2. A practicable model for the EPS system has been
strategy to improve steering maneuverability on slippery
established, including the full vehicle mechanical roads. SAE, 2002(1).
system, the EPS mechanical system, and the EPS 9 Kellner, M., Neumann, M., Banerjee, A., and Doshi, P.
electric control system. Parametrization of modelica models on PC and real
3. The simulation model has been correlated with test time platforms. In (Ed. C. Kral) Proceedings of the 5th
data using the HiL method and dSPACE hardware International Modelica Conference, Vienna, Austria, 4–5
and software. September 2006, pp. 267–273.

JMES925 © IMechE 2008 Proc. IMechE Vol. 222 Part C: J. Mechanical Engineering Science

You might also like