Professional Documents
Culture Documents
fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TMECH.2018.2812220,
IEEE/ASME Transactions on Mechatronics
1083-4435 (c) 2018 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TMECH.2018.2812220,
IEEE/ASME Transactions on Mechatronics
can effectively improve the vehicle handling [11]. Russell et al. mvx ( & r ) FYf cos f FYr cos r
proposed a control strategy based on the combination of linear (1)
and nonlinear state feedback with feedforward of reference I z r& FYf a cos f FYr b cos r
model yaw moment, lateral force, and longitudinal force at the here m is the mass of vehicle; Iz is the yaw moment of inertia;
vehicle center of gravity. Experimental results demonstrate β is the vehicle side slip angle; r is the vehicle yaw rate; vx is
successful emulation of the low friction reference model’s the longitudinal velocity of vehicle; a and b are respective
planar dynamics for a variety of driving maneuvers and distances from the center of gravity to the front and rear axles;
reference friction coefficients [12]. Meanwhile, Russell et al. δf and δr are respective the steering angles of front and rear
proposed a forcebased approach to emulating the lateral wheel; Fyf and Fyr are respective lateral tire forces of front and
dynamics of a vehicle on a low friction surface using four- rear tires.
wheel steer-by-wire. The steer angle commands are computed Assuming that there is no longitudinal load transfer, the
with a combination of feedforward and state feedback, and the vehicle weight is distributed among the front and rear axle
controller explicitly handles rear wheel actuator limitations. based on the vehicle dimension.
Experimental results from an asphalt surface confirm this b
controller successfully tracks the lateral dynamics of the low Fzf a b mg
friction reference model [13]. However, the robustness and (2)
robust performance of the system with feedforward and F a mg
zr
feedback hybrid control are not analyzed. And the feedforward ab
and feedback hybrid controller can’t play MIMO advantage of here Fzf and Fzr are respective vertical forces of front and rear
FSV [14]. axles.
According to the law of variable transmission ratio Tire is an important part of automobile. The forces of
derived from constant yaw rate gain, reference model is automobile are mostly from the contact between tires and road
established to make the vehicle have good steering surface, so the tire model plays an important role in the
characteristics. The parameters of the steering system are simulation of vehicle handling stability.
perturbed with the change of vehicle’s running condition. In The tire slip angles are modeled as kinematic functions of
addition, the external disturbances can also cause dynamic the states and the steer angles:
perturbations. Therefore, the proposed controller should have a
strong robustness and robust stability. Meanwhile, FSV is a f v r f
x
MIMO system. Aiming at the MIMO system and considering (3)
b
the parameters uncertainties of vehicle speed and tire r
r vx
r
cornering stiffness, the structured singular value μ is used to
analyze FSV under multiple perturbations in this paper. Based here αf and αr are the slip angles of front and rear tire.
on μ control strategy, μ controller is designed to track the In general, tire forces are nonlinear with slip angles.
desired sideslip angle and yaw rate and thus vehicle gets Considering follow-up control strategies, the linear tire model
vehicle performance [15]. is established. The nonlinearity of tire is considered as
The remainder of this paper will describe the yaw and parameters uncertainty.
lateral stability control and simulation and experiment results. Fyf Caf f
Firstly, the system model and reference model are described in (4)
Section II. Section III develops the details about stability Fyr Car r
control, including the solution of weighted function and design here Caf and Car are the cornering stiffness of front and rear
of μ controller. Simulation results and analysis are presented wheel.
in Section IV. In Section V, the experiment results are showed ì
ïC af = Caf (1 + p1 d1 )
to validate the simulation results. Finally, conclusions are ï
í (5)
discussed in Section VI. ï
îC ar = C ar (1 + p2 d2 )
ï
ï
II. SYSTEM MODELING Caf and Car are the nominal values of cornering stiffness of
To accomplish further control strategy design and front and rear wheel; p1 and p2 are the ranges of perturbations;
simulation, FSV dynamics model including vehicle model, the d1 and d2 are the standard perturbations.
steer-by-wire system model and reference model is established.
B. Four-wheel steer-by-wire system
A. Four-wheel steer-by-wire vehicle dynamics model
The structure of steering actuator is shown in Fig.1. FSV
For FSV, lateral dynamics [16] is very important and it cancels the mechanical connection between the steering wheel
can reflect the steering characteristics of vehicle. Meanwhile, and the wheel. According to the steering signal from steering
longitudinal forces are not modeled. FSV model is established wheel sensor, ECU gives the commands to the steering
neglecting the vehicle rolling and only considering the lateral motor[17].
and yaw two degrees of freedom. The dynamic model of FSV
can be expressed as:
1083-4435 (c) 2018 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TMECH.2018.2812220,
IEEE/ASME Transactions on Mechatronics
Re
driver's manipulation of vehicles. This is the variable
transmission ratio derived from constant yaw rate gain.
Le ie According to the law of variable transmission ratio
derived from constant yaw rate gain, the relationship between
steering wheel angle and front wheel steering angle is:
sw u/L
is (10)
f Gsw (1 Ku2 )
where L is wheelbase of the vehicle; Gsw is the constant yaw
xr
mr m a b
rate gain; K is stability factor, K 2 ( ).
Br
Br L Car Caf
Mr Mr The steady-state yaw rate r* is a constant and now & and
r& both are zero. The relationship between desired yaw rate
and front wheel steering angle is:
r* u/L
Figure.1 Steering actuator (11)
f 1 Ku2
Electrical equation of steering motor can be modeled as:
The ideal sideslip roll angle can be depicted as:
Li& Ri E U
(6) * 0 (12)
E K b&m
where L is the steering motor inductance; R is steering motor III.STABILITY CONTROL
resistance; E is steering motor back EMF; U is the voltage of
the motor; Kb steering motor back EMF constant. Based on the established FSV model, simple feedforward
The differential equation can be modeled as: and feedback control strategy is applied to verify the
advantage of FSV. The structure of feedforward and feedback
J m&&m Bm&m Ta Tm control strategy is shown in Fig.2.
(7)
Tm kt i
where Jm is the moment of inertia of steering motor; Bm is
damping coefficient of steering motor; Tm is electromagnetic
torque of steering motor; i is electric current of steering motor;
Kt is electromagnetic torque constant; θm is steering angle of
steering motor; Ta is output torque of steering motor.
Differential equations of motion of rack can be modeled
as:
M r &x&r Br x&r FR NTa / rp
(8)
sg xr / rp
where Mr is the mass of rack; Br is damping coefficient of rack; Figure.2 Structure of
FR is steering resistance of rack; N is reduction ratio of the feedforward and feedback control strategy
steering motor; xr is displacement of the rack; rp is radius of As is shown in Fig.2, K is SBW controller. According to
the gear; θsg is steering angle of the gear. the controller of front wheel, steering command is obtained.
The input of rear wheel is get on the base of C1 and C2. They
C. Dynamics of the reference model are feedback control coefficient. The rear wheel angle can be
The reference model [18] is described by using the single expressed as the following equation.
track vehicle model. FWS vehicle can be modeled as: r C1 f C2ux r (13)
C af a C ar b
mv x ( r ) (C af C ar ) r C af f
& Steering wheel angle step simulation results of FSV,
vx SBW vehicle and 4WS vehicle are shown in Fig.3 and Fig.4.
(9)
The steering wheel angle is 90deg, and the speed is 25m/s.
& C af a 2 Car b 2
I z r (C af a C ar b ) vx
r C af a f
Steering wheel is operated to vehicle by driver according
to the deviation between current position and desired position.
Assuming that we can make the vehicle maintain a definite
and predictable relationship between running angle and
steering wheel angle at different speed, it will greatly facilitate
1083-4435 (c) 2018 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TMECH.2018.2812220,
IEEE/ASME Transactions on Mechatronics
-0.06
0 1 2 3 4 5
time(t/s)
0.5
0.4
yaw rate(r/(rad/s))
0.3
0.2 ideal r
FSV
0.1 SBW Figure.5 Simulation block
4WS diagram of four-wheel steer-by-wire vehicle
0 As is shown in Fig.5, FSV received operation inputs
0 1 2 3 4 5
time(t/s)
from driver and then road feeling motor is used to simulate
sense of road. At the same time, according to the ECU signals
Figure.4 Yaw rate response
transmitted from the sensors (steering wheel angle and vehicle
As can be seen from Fig.3, the sideslip angle peak value
speed) and the reference vehicle model, reference states (ideal
of FSV is 0.006rad. It’s 0.01rad for 4WS vehicle. While for
sideslip angle and the ideal yaw rate) are obtained by
SBW system, the sideslip angle peak is 0.058rad. So
calculation. The deviations between the reference states and
compared with 4WS vehicle, the sideslip angle peak value of
the actual states (sideslip angle and yaw rate) are the inputs of
FSV is reduced by 40% and as for SBW vehicle, it’s 90%.
μ controller. And then ideal front and rear wheel angles are got
Fig.4 is the yaw rate response of FSV, SBW vehicle and 4WS
from controller. The ideal front and rear wheel angles are
vehicle. It can be seen from Fig.4 that FSV and SBW vehicle
converted into the actuator command.
can well track the desired yaw rate. Meanwhile, the response
speed of FSV is 0.2s faster than SBW vehicle and the A. Uncertainty representation
overshoot of FSV is 8% less than SBW vehicle. 4WS vehicle In section 2, nonlinear vehicle model is simplified as
can’t track the desired yaw rate. The FSV can make the linear model and the nonlinearity of tire is considered as
vehicle get desired sideslip angle and yaw rate by two active parameters uncertainty. Moreover, the vehicle speed vx varies
inputs. Moreover, the vehicle is subjected to external factors, with the process of turning.
such as lateral wind disturbance, road disturbance, sensor The vehicle dynamics model can be described by the
noise and parameter perturbation of the steering system itself, following transfer matrix in the Laplace domain:
which always result in deterioration and instability of the
system performance. So, to ensure that the controlled object is
G f (s) Gr (s) f
r G ( s) G (s) (14)
running in accordance with the intended target, the system fr rr r
must maintain the robust stability for the uncertainty. Besides, Parametric uncertainty can be expressed as complex
the system needs to have good performance indicator, that is, uncertainties to simplify the controller design. The uncertainty
robust performance. The feedforward and feedback integrated region can be expressed as multiplicative uncertainty:
control strategy can’t guarantee the robust stability and robust G p ( s ) G ( s )(1 W I I ( s )) (15)
performance of the system.
Multiplicative weight function is derived from:
FSV combines with SBW vehicle and 4WS vehicle. The
sideslip angle and yaw rate are the two important state G p ( j ) G( j )
I ( s) max , (16)
parameters. The sideslip angle describes the path tracking G ( j )
ability and yaw rate reflects the vehicle stability[19]. The main
1083-4435 (c) 2018 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TMECH.2018.2812220,
IEEE/ASME Transactions on Mechatronics
0.5
Bode Diagram
1
Magnitude (abs)
0.4
0.9
0.8 0.3
0.7
0.2
Magnitude (abs)
0.6
0.5 0.1
0.4
0
-4 -2 0 2
0.3 10 10 10 10
Frequency (Hz)
0.2
d
0.1
Figure.6 Multiplicative uncertainty weighted functions. (a), (b),
0
-3 -2 -1 0 1 2 3 (c) and (d) are the multiplicative uncertainty weighted
10 10 10 10 10 10 10
Frequency (Hz) functions of Gfβ(s), Gfr(s), Grβ(s) and Grr(s). The dotted line is
a the relative error and solid line is the weighted functions.
As is shown in Fig.6, the dotted lines are the relative
Bode Diagram error between nominal object and uncertain object. Solid lines
0.7
are the weighted functions. The relative errors of Gfβ(s) and
0.6 Gfr(s) are shown in Fig.6.a and Fig.6.c. It can be seen that the
peak value of low-frequency region is 0.99 and the peak value
0.5
of high-frequency region is 0.22. Similarly, for Grβ(s) and
Grr(s), the peak value of low-frequency region is 0.545 and the
Magnitude (abs)
0.4
peak value of high-frequency region is 0.15. Therefore, the
0.3
weighted function matrix is selected as
é 0.1s + 0.99 ù
0.2
ê 0 ú
ê 0.1 ú
s+1
éW 0 ù êê 0.22 ú
0.1
WI = ê f ú=ê ú
ê0 ú (17)
0
ë Wr úû ê 0.06s + 0.545 ú
0
-3 -2 -1 0 1 2 3
10 10 10 10 10 10 10
ê 0.06 ú
Frequency (Hz)
ê s + 1 úú
b êë 0.15 û
Bode Diagram
Wp is the performance weighted function. In this paper,
1 the integral function is used to fit the function.
0.9
s 7.5
0.8 1.5s 0.00075 0
Wp (18)
0.7
s 7.5
0
1.5s 0.00075
Magnitude (abs)
0.6
0.5
1083-4435 (c) 2018 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TMECH.2018.2812220,
IEEE/ASME Transactions on Mechatronics
C. μ controller solution
Considering any commutative matrix with Δ, then
( DM ) (MD)
(27)
( DMD ) (M )
1
iteration 3
perturbation Δ are stable. Consequently, the sufficient and 1.2
iteration 4
necessary condition for the robust stability is turned to: 1.1
( M ( j )) 1, (23) 1
When all the sub blocks are complex perturbation,
0.9
structured singular value can be expressed as:
( N ) m ax , ( ) 1 ( N ) (24) 0.8
1083-4435 (c) 2018 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TMECH.2018.2812220,
IEEE/ASME Transactions on Mechatronics
yaw rate(r/(rad/s))
0.46
0.3
0.44
Parameter value unit 0.42
Mr 2 Kg 0.2 0.8 1 1.2 1.4 1.6
Br 90 N/m ideal r
rp 7.367×10-3 m 0.1 μ control r
N 20 -- hinf control r
Jm 9.8×10-4 Kg·m2 0
0 1 2 3 4 5
Kt 0.65 N m/ A time(t/s)
Kb 0.0345 V /(rad / s) b. Yaw rate
L 2.69×10-4 H Figure.10 The simulation
R 0.21 Ω results of steering wheel angle step input.
m 1653 Kg It can be seen from Fig.10.a that the sideslip angle of
a 1.463 m FSV is within 0.008rad with μ control. While with H∞ control
b 1.585 m it is within 0.012rad which is 50% higher than that with μ
Iz 2765 Kg·m2 control. Therefore, the risk of sideslip with μ control decreases.
In order to verify the tracking performance for reference Fig.10.b is yaw rate response of FSV. As shown in Fig.10.b,
model of designed controller and the vehicle performance of FSV with μ control and H∞ control both can well track the
research object, steering wheel angle step input and double ideal yaw rate. But the overshoot of yaw rate with μ control is
lane change simulation experiments in which vehicle runs on 5.8% and the overshoot of yaw rate with H∞ control is 6.4%.
good road at high speed and runs on low adhesion coefficient Therefore, the steady-state response of the FSV controlled by
road at low speed are designed. Moreover, simulation results μ is better than that controlled by H∞.
with μ control and H∞ control two kinds of control algorithms In order to study the influence of the four-wheel steer-by-
are compared to verify the μ control performance. The wire system on the vehicle dynamic characteristics more
simulation results of steering wheel angel step input are shown comprehensively, double lane change simulation experiments
in Fig.10. In the simulation experiment, the steering wheel in which vehicle runs on good road at high speed and runs on
angle is 90deg, and the speed is 25m/s low adhesion coefficient road at low speed are designed.
Fig.11 is the double lane change simulation result. In the
0.01
simulation experiments, the speed is 25m/s and μf is 0.9.
0.005 0.04
sideslip angle(β/rad)
0 0.02
sideslip angle(β/rad)
-0.005 0
-0.01 -0.02
μ control
hinf control
-0.04
-0.015
0 1 2 3 4 5 μ control
time(t/s) hinf control
-0.06
a. Sideslip angle 0 1 2 3 4 5
time(t/s)
a. Sideslip angle
1083-4435 (c) 2018 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TMECH.2018.2812220,
IEEE/ASME Transactions on Mechatronics
are shown in Fig.11.c. They are basically the same with two
1.5
kinds of control. Because vehicle can’t quickly return, front
1 and rear wheel angles are fluctuant at the end of double lane
0.5
change. Fig.11.d is trajectory tracking of double lane change.
It can be seen from Fig.11.d that trajectory tracking
yaw rate(r/(rad/s))
0
performance of vehicle with μ control is better than that with
-0.5 H∞ control. The simulation results show that the
-1
-0.6 maneuverability of FSV under high speed with μ control is
-0.8 μ ideal r better than that with H∞ control.
-1.5 μ control r Double lane change simulation experiments in which
-1
-2 hinf ideal r vehicle runs on low adhesion coefficient road (μf=0.2) at low
2 2.2 2.4 2.6 2.8 hinf control r speed are carried to reflect the control effect. The simulation
-2.5
0 1 2 3 4 5 results are shown in Fig.12
time(t/s)
0.02
b. Yaw rate
0.015
0.4
0.01
sideslip angle(β/rad)
0.3
0.005
0.2
0
wheel angle(δ/rad)
0.1
-0.005
0
-0.01
-0.1 μ control
-0.015
μ control δf hinf control
-0.2 -0.02
μ control δr 0 2 4 6 8
-0.3 hinf control δf time(t/s)
hinf control δr
-0.4 a. Sideslip angle
0 1 2 3 4 5
time(t/s)
5
yaw rate(r/(rad/s))
target path
0
4 μ path
hinf path
lateral displacement(y/m)
3 -0.5
-0.5 μ ideal r
2 -1 -0.6 μ control r
-0.7 hinf ideal r
1 hinf control r
-1.5 2.5 3 3.5
0 2 4 6 8
0 time(t/s)
-1
b. Yaw rate
0 10 20 30 40 50 60 70 80
longitudinal displacement(x/m) 0.4
0.1
As shown in Fig.11.a, it is double lane change road
sideslip angle simulation result. The sideslip angle is less than 0
0.02rad with μ control while it is within 0.04rad with H∞ -0.1
control. Compared with H∞ control, μ control makes the μ control δf
-0.2
sideslip angle of vehicle reduce 50%. It can be seen from μ control δr
Fig.11.b that FSV with μ control and H∞ control can well track -0.3 hinf control δf
hinf control δr
the ideal yaw rate, but at the end of the double lane change, -0.4
0 1 2 3 4 5 6 7 8
vehicle with H∞ control can’t quickly return. The vehicle with
time(t/s)
μ control can well track yaw rate. Front and rear wheel angles
c. Wheel angles
1083-4435 (c) 2018 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TMECH.2018.2812220,
IEEE/ASME Transactions on Mechatronics
5
target path
4 μ path
hinf path
lateral displacement(y/m)
-1
0 10 20 30 40 50 60 70
longitudinal displacement(x/m)
than that with H∞ control. It can be seen from Fig.12.d that the
vehicle with μ control and with H∞ control both can well track 6
the target path. The simulation results show that the control
effect of μ is better than that of H∞. 4
Through analysis for the results of steering wheel angle
step input and double lane change simulations in which 2
vehicle runs on good road at high speed and runs on low
adhesion coefficient road at low speed, μ controller can make 0
the FSV well track the reference model. Therefore, the risk of
sideslip is reduced. Meanwhile, it can make vehicle have -2
0 5 10 15 20 25
better maneuverability by good tacking performance to yaw
time t (s)
rate.
a. Longitudinal velocity
V. EXPERIMENT VALIDATION 0.7
In order to further verify the control effect of the
0.6
designed μ controller, the vehicle experiment is carried out.
sideslip angle β (deg)
1083-4435 (c) 2018 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TMECH.2018.2812220,
IEEE/ASME Transactions on Mechatronics
10
10 15
reference r
reference r
10 actual r
actual r
5
5
0
0
-5
-5
-10 -10
0 5 10 15 20 25 0 10 20 30 40
time t (s) time t (s)
c. Yaw rate c. Yaw rate
Figure.14 Experiment results under double lane change test. Figure.15 Experiment results under serpentine line test.
6 ACKNOWLEDGEMENT
This study was supported by the National Natural Science
5
0 10 20 30 40 Foundation of China (No.51775268), National Key R&D
time t (s) Program of China (No. 2017YFB0103604) and the
a. Longitudinal velocity Fundamental Research Funds for the Central Universities (No.
NE2016002).
2.5
2 REFERENCES
[1] Sande T V D, Zegelaar P, Besselink I, et al. A robust control analysis for
sideslip angle β (deg)
1083-4435 (c) 2018 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TMECH.2018.2812220,
IEEE/ASME Transactions on Mechatronics
11
Vehicle[J]. IEEE Transactions on Industrial Electronics, 2016, Xiaoxi Qin is currently a graduate
63(9):5774-5785.
student in the Department of Vehicle
[6] Zhang H, Zhao W Z. Stability control strategy of steer-by-wire system
based on LQG/LTR[J]. Science China, Technological Sciences, Engineering, Nanjing University of
2017(6):1-10. Aeronautics and Astronautics, Nanjing,
[7] Canale M, Fagiano L. Stability control of 4WS vehicles using robust China. His research interests include
IMC techniques[J]. Vehicle System Dynamics, 2008, 46(11):991-1011.
vehicle system dynamics.
[8] Li M X, Jia Y M and Du J P. LPV control with decoupling performance
of 4WS vehicles under velocity-varying motion[J]. IEEE transactions on
control systems technology, 2014, 22: 1708-1724.
[9] Li M X, Jia Y M. Decoupling control in velocity-varying four-wheel Chunyan Wang received the B.S. and
steering vehicles with H-infinity performance by longitudinal velocity
Ph.D. degrees in Mechanical
and yaw rate feedback[J]. Vehicle System Dynamics: International
Journal of Vehicle Mechanics and Mobility, 2014, 52: 1563-1583. engineering from Jilin University,
[10] Hang P, Chen X B, Fang S D, et al. Robust control for four-wheel- Changchun, China, in 2000 and 2008,
independent- steering electric vehicle with steer-by-wire system[J]. respectively. She is currently an
International Journal of Automotive Technology, 2017, 18(5):785-797.
associate professor in the Department
[11] Zhu T, Zong C. Advanced Integrated Control Strategy Based on SBW
and 4WS[C]. Pacific-Asia Conference on Circuits, Communications and of Vehicle Engineering, Nanjing
Systems. IEEE Computer Society, 2009:512-515. University of Aeronautics and Astronautics, Nanjing, China.
[12] Russell H E B, Gerdes J C. Design of Variable Vehicle Handling Her research interests include vehicle system dynamics.
Characteristics Using Four-Wheel Steer-by-Wire[J]. IEEE Transactions
on Control Systems Technology, 2016, 24(5):1529-1540.
[13] Russell H E B, Gerdes J C. Low friction emulation of lateral vehicle
dynamics using four-wheel steer-by-wire[J]. American Control
Conference, Jun. 2014:3924-3929.
[14] Lian Y F, Wang X Y, Zhao Y, et al. Direct Yaw-moment Robust
Control for Electric Vehicles Based on Simplified Lateral Tire Dynamic
Models and Vehicle Model[J]. IFAC PapersOnLine, 2015, 48(28):33-38.
[15] Hu J S, Wang Y, Fujimoto H, et al. Robust Yaw Stability Control for In-
wheel Motor Electric Vehicles[J]. IEEE/ASME Transactions on
Mechatronics, 2017, 22(3):1360-1370.
[16] Pacejka H B. Tire and Vehicle Dynamics[J]. Butterworth Heinemann,
2012.
[17] Lv C, Liu Y, Hu X, et al. Simultaneous Observation of Hybrid States for
Cyber-Physical Systems: A Case Study of Electric Vehicle
Powertrain[J]. IEEE Transactions on Cybernetics, 2017, PP(99):1-11.
[18] M. Mirzaei and H. Mirzaeinejad. Fuzzy Scheduled Optimal Control of
Integrated Vehicle Braking and Steering Systems, IEEE/ASME
Transactions on Mechatronics,2017, PP(99):1-1.
[19] Lv C, Xing Y, Zhang J, et al. Levenberg-Marquardt Backpropagation
Training of Multilayer Neural Networks for State Estimation of A Safety
Critical Cyber-Physical System[J]. IEEE Transactions on Industrial
Informatics, 2017, PP(99):1-1.
[20] Nwesaty W, Bratcu A I, Sename O. Power sources coordination through
multivariable linear parameter-varying H-infinity control with
application to multi-source electric vehicles[J]. Iet Control Theory &
Applications, 2016, 10(16):2049-2059.
[21] Wang C Y, Deng K, Zhao W Z, et al. Stability control of steer by wire
system based on μ synthesis robust control[J]. Science China,
Technological Sciences, 2017, 60(1):16-26.
1083-4435 (c) 2018 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.