International Journal of Automotive Technology, Vol. 11, No. 2, pp.

205213 (2010)

DOI 10.1007/s1223901000260

Copyright 2010 KSAE

12299138/2010/05108

IMPROVEMENT OF DRIVABILITY AND FUEL ECONOMY

WITH A HYBRID ANTISKID BRAKING SYSTEM
IN HYBRID ELECTRIC VEHICLES
J. L. ZHANG , CH. L. YIN and J. W. ZHANG
*

School of Mechanical Engineering, Shanghai Jiao Tong University, Shanghai 200240, China
(Received 20 February 2008; Revised 15 December 2008)

ABSTRACTWhen braking on wet roads, Antilock Braking System (ABS) control can be triggered because the available

brake torque is not sufficient. When the ABS system is active, for a hybrid electric vehicle, the regenerative brake is switched
off to safeguard the normal ABS function. When the ABS control is terminated, it would be favorable to reactivate the
regenerative brake. However, recurring cycles from ABS to motor regenerative braking could occur. This condition is felt to
be unpleasant by the driver and has adverse effects on driving stability. In this paper, a novel hybrid antiskid braking system
using fuzzy logic is proposed for a hybrid electric vehicle that has a regenerative braking system operatively connected to an
electric traction motor and a separate hydraulic braking system. This control strategy and the method for coordination between
regenerative and hydraulic braking are developed. The motor regenerative braking controller is designed. Control of
regenerative and hydraulic braking force distribution is investigated. The simulation and experimental results show that
vehicle braking performance and fuel economy can be improved and the proposed control strategy and method are effective
and robust.

KEY WORDS : Hybrid antiskid braking System, Fuzzy controller, Hybrid electric vehicle

1. INTRODUCTION

applying supplemental hydraulic braking only when the

braking torque supplied by the electric traction motor
cannot meet the driver's brake demand. The amount of
kinetic energy that is wastefully dissipated while driving or
launching an electric or hybrid vehicle is decreased if
energy losses due to wheel slippage can be minimized.
During deceleration, the electric motor takes the regenerative braking torque from the powertrain and the kinetic
energy of the vehicle can be regenerated into the battery as
electrical energy. Therefore, a hybrid electric vehicle has
two different braking torques: one is the regenerative braking torque that is offered by the electric motor; the other is
the hydraulic braking torque that is provided by the hydraulic braking system. When the motor is only connected to
the front axle, the regenerative braking torque can only act
on the front axle and the front wheels can easily slip and
lock when the roads friction coefficient is low. Because of
the regenerative braking torque, the hydraulic braking
torque distribution between the front and rear axles should
be changed compared with traditional vehicles. Therefore,
the logic underlying the control of the Antilock Braking
System (ABS) control system ought to be adjusted to
account for the regenerative braking system.
Fuzzy logic is a branch of logic that uses degrees of
membership, rather than strict true/false or binary membership, to define a set. As a result, fuzzy sets have nondefinite
boundaries. The degree of membership represents the level

The focus of current research on electric and hybrid electric

vehicles has been on increasing energy efficiency and
reducing emissions (Oh et al., 2005). Antiskid braking
systems are well known and commonly employed in conventional internal combustion engine vehicles. However, the
development of electric and hybrid vehicles has introduced
concerns and opportunities unique to electric vehicle control system design. A primary area of concern in electric
vehicle design, because of limited battery storage, is the
amount of energy used by the various vehicle subsystems.
Therefore, designing vehicle control systems that minimize
energy use and conserve the available energy stored in the
battery is critical.
Another opportunity unique to the electric or hybrid
vehicle is the possibility of regenerating the kinetic energy
dissipated during braking, or any other period in which the
accelerator pedal is not depressed and the vehicle is in
motion, e.g. coasting. Such regeneration can be accomplished by controlling the operation of the electric traction
motor so that it behaves like a generator. The kinetic energy
recovered during this process can be used to recharge the
traction battery and stored for future use (Jung et al., 2007).
The amount of energy recovered can be increased by
*Corresponding author. e-mail: zjlong@yahoo.cn
205

206

J. L. ZHANG, CH. L. YIN and J. W. ZHANG

of confidence that a particular value belongs in a given

fuzzy set. David Elting and Mohammed Fennich reported
that automotive systems realize superior characteristics
through the use of fuzzy logic controllers, especially in
nonlinear cases (David et al., 1994). The brake system is
a challenging control problem because vehicle brake
dynamics are highly nonlinear with uncertain time-varying
parameters. Fuzzy controllers have the benefit of not
requiring a mathematical model of the plant. Also, certain
fuzzy control designs can be implemented that have the
ability to learn or to adapt themselves to improve their
performance. Because of these features, in recent years,
fuzzy logic control techniques have been applied to a wide
range of systems (Benbouzid et al., 2006). Many electronic
control systems in the automotive industry, such as automatic transmissions (Zhao et al., 2007 and Baumann et al.,
2000) and antiskid brake controls (Yu et al., 2002 and
Mokhiamar and Abe, 2006), are currently being pursued.
These electronically controlled automotive systems gain
superior characteristics through the use of fuzzy logic.
Antiskid brake control systems for a conventional vehicle
based on fuzzy inference are presented in many papers.
However, the control systems presented in those papers
were designed solely for operation in conventional vehicles
having internal combustion engines and, as such, do not
provide for coordinated control of regenerative braking and
antiskid braking (Hybrid Antiskid Braking System) in
hybrid electric vehicles.
Accordingly, there is a need for a control method for
providing a Hybrid Antiskid Braking System for hybrid
electric vehicles that improves braking performance, provides
maximum regenerated kinetic energy during braking, minimizes the loss of kinetic energy and coordinates hydraulic
braking torque and regenerative braking torque with each
other.

2. HYBRID ANTISKID BRAKING SYSTEM

The hybrid antiskid braking system includes at least two
driven wheels, a regenerative braking system operatively
connected to an electric traction motor having motor control means for controlling the operation of the motor, and a
separate hydraulic braking system having brake control
means for adjusting the fluid pressure applied at each
wheel by the hydraulic braking system.
The method for providing hybrid antiskid braking control comprises the steps of: sensing vehicle parameters to
obtain real-time measurements defining vehicle behavior;
calculating additional vehicle parameters describing vehicle
behavior not directly measured in the step of sensing vehicle parameters; determining whether regenerative braking
control or hydraulic braking control is required based on
the vehicle behavior and the critical values; applying fuzzy
logic to the vehicle parameters; generating command signals
in response to the vehicle parameters and the required
control; and controlling the motor control means of the

Figure 1. Ground-fixed and vehicle-fixed reference frames,

G and A.
electric traction motor and the brake control means of the
hydraulic braking system in accordance with the command
signals.
2.1. Vehicle Model
In order to develop the equations of motion for the basic
model, a suitable reference frame must be defined. In the
vehicle-fixed axis system, A, (see Figure 1) the vehicle has
the following velocity components:
Forward speed, u, in the x direction,
Lateral velocity, v, in the y direction,
Yaw velocity, Wr, in the z direction.
(

du/duv Wr )G/ g=( Fx11 + Fx12 )cosd

( F y11 + F y12 )sin d Ff F w

(1)

dv/ du+u Wr )G/g=( Fy11 + Fy12)cos d+

( Fx11 + F x12 )sin d Fy21 F y22
Iz dWr/dt=( Fy11 + Fy12 )cosd a( Fy21 + Fy22 )b+
( Fx11 + F x12 )sin d a ( F x11 F x12 )cos d B f /2+
( Fy11 F y12 )sin d B f /2
(

(2)

(3)
(4)
Fw Cd Av
2
(5)
Ff Gf Gf0 v
(6)
IwdW(1,1)/dt = Fx RwMf (1,1)Thydra (1,1)Tmotor reg/2
IwdW(1,2)/dt = Fx RwMf (1,2)Thydra (1,2)Tmotor reg/2
(7)
IwdW(2,1)/dt = Fx RwMf (2,1)Thydra (2,1)
(8)
IwdW(2,2)/dt = Fx RwMf (2,2)Thydra (2,2)
(9)
in which Fx is the tire longitudinal force, Fy is the tire lateral
force, Ff represents the rolling resistance, Bf is the distance
between two front tires, Fw denotes the windward resistance, Iz denotes its yaw inertia about the mass center, Iw
is the tire moment of inertia, A is the windward area, a is
the distance between the center of mass and the front axis,
b is the distance between the center of mass and the rear
axis, Rw is the tire rolling radius, G is the vehicle weight, g
is the acceleration due to gravity, d is the front steer angle,
Mf (1,1)~Mf (2,2) are the wheel rolling resistance torques,
2

/21.15
(1 +

/19440 )

11

12

21
22

IMPROVEMENT OF DRIVABILITY AND FUEL ECONOMY WITH A HYBRID ANTISKID BRAKING SYSTEM 207
(1,1)~ (2,2) are the wheel yaw velocities, motor_reg is the
motor regenerative braking torque, hydra is the frictional
brake torque, d is the windward damping coefficient, and
is the coefficient of friction.

2.2. Motor Model

Most HEVs employ both a conventional braking system
and a regenerative braking system. The conventional braking system typically includes frictional drum or disc braking assemblies selectively actuated by a hydraulic system.
The regenerative braking system utilizes the electric motor,
providing negative torque to the driven wheels and converting kinetic energy to electrical energy for recharging the
battery or power supply. Modern electric drive motors are
sophisticated systems with microprocessor-based controllers,
advanced power electronics, and sophisticated control
algorithms. Figure 2 shows numerical modeling of the
electric motor based on experimental data.
During motor regenerative braking, there are many
factors that influence regenerative braking torque, such as
battery state of charge (SOC), motor speed and motor
temperature. More details are presented in section 2.7 of
this paper.
2.3. Tire Model
The modeling of tire force plays an important role in

determining the vehicles dynamic behavior. Generally, tire

modeling can be categorized into two different approaches.
One is the development of an analytical model (Sakai,
1990); the other is the construction of a semi-empirical
model (Pacejka, 2002). The latest semi-empirical tire
model, the Magic Formula by Pacejka, is used in this study
for its convenience and accuracy.
An important quantity for the tire friction force calculation is the wheel vertical load, zij, which is a function of
both the vehicles static load and the dynamic load transfer.
It can be represented as:
mgb
h ma bh K + D -
(10)
F =---------- ma ------ --------------- -----------------------2L
2L LT
T

mgb
h ma bh K + D (11)
F =---------- ma ------ + --------------- -----------------------T
2L LT
2L
mga
h ma ah K + D -
(12)
F =---------- +ma ------ --------------- ------------------------2L
2L LT
T

mga
h ma ah K + D (13)
F = ----------+ma ------ +--------------- +------------------------T
2L
2L LT
Wherein x and y are the vehicle longitudinal and lateral
accelerations. zfl , zfr , zrl and zrr are the wheel vertical
load for the front left, front right, rear left and rear right
wheels respectively. Equations (10-13) are instantaneous
relationships between accelerations and vertical forces. No
suspension effects are considered.
Tire lateral force calculation depends on wheel vertical
load and slip angle, and the slip angle of each wheel is
calculated as:
v + aW
(14)
a =darctan -----------------------------
u 0.5T W
v + aW -
(15)
a =darctan --------------------------u+0.5T W
F

zfl

zfr

zrl

zrr

fl

fr

vbW
a = arctan -----------------------------
u0.5T W

(16)

vbW -
a = arctan ---------------------------u0.5T W

(17)

rl

rl

Moreover, the longitudinal wheel slip is defined as:

R u
ij = ----w-------ij----------ij , Rw ij < uij

(18)

R u
ij = ----w-------ij----------ij , Rw ij uij
Rw ij

(19)

uij

Figure 2. Torque and efficiency characteristics of a motor

based on the testing data.

In this paper, driving state is not considered, so the longitudinal wheel slip is that of the braking state in Equation
(18).
An accurate determination of the adhesion coefficient,
d, is critical for determining the current operating point
of a driven wheel. The present application calculates d
for each driven wheel according to the following equations:

208

J. L. ZHANG, CH. L. YIN and J. W. ZHANG

d = [ k* + * 2d 2Jrd] /2RwNv

(20)
*
(21)
k k( (e Gr )rdld)
*
(22)
( ( e Gr ) rd ld )
wherein k is the spring rate of the vehicles drive axle, e,
is the integral of the electric motor speed, Gr is the combined gear ratio of the differential and the transmission, rd
and ld are the integrals of the speed of the right and left
driven wheels respectively, Jr is the wheel inertia, is the
damping rate between the motor shaft and wheel due to
bearings, e is the electric motor speed, rd is the speed of
the right driven wheel, ld is the speed of the left driven
wheel, d is the time derivative of the speed of the driven
wheel being monitored, i.e. the driven wheels acceleration
rate, d is the hydraulic brake torque applied to the
monitored driven wheel, Rw is the rolling radius of the
wheel, and Nv is the normal force on the respective wheel.
=

2.4. Coordination Method for a Regenerative and Antiskid

Braking System
On a surface with a low coefficient of friction, such as ice,
ABS control can be triggered because the required brake
torque is not sufficient. When the ABS system is active, the
regenerative brake is switched off to uncouple the wheels
from the driving track and safeguard the normal ABS
function.
When the ABS control is terminated, it would be favorable to readmit the regenerative brake. However, this would
shift the total braking power to the driving axle again, and
permanent cycling (cyclic change-over ABS regenerative
brake ABS etc.) could occur. This condition is felt to
be unpleasant by the driver and has adverse effects on
driving stability (Jochen and Gregor, 2006).
Thus, if someone attempts to apply the antiskid method
with motor control, cooperation or interference between
regenerative and hydraulic braking must be considered (see
Figure 3).
Items 1 to 6 mentioned in the text correspond to
conditions 1 to 5 listed in Figure 3.
1. In the partial braking range (before the entry into ABS
control), the demanded braking power is realized by means
of the regenerative brake up to a maximum value determined by the generator; any additional demand is realized

Figure 3. Coordination between regenerative braking and

antiskid braking.

by means of the wheel brake.

2. The regenerative brake is deactivated upon entry into
ABS control. The demanded braking power is completely
realized by means of the wheel brake (in ABS control).
3. After ABS control ceases, a predetermined time elapses.
The regenerative brake remains disconnected (as in 2) during this time. The braking demand is realized only by the
wheel brake. The pressure in the wheel brake cylinders at
the end of ABS control is determined, and the corresponding brake torque is calculated and stored by the control
system.
4. The regenerative brake is switched on again upon
expiry of 3. However, the brake torque is limited. The limit
is raised with a preset gradient starting from 0 until a
defined portion of the brake torque of the driving axle
determined in 3 has been reached at the end of the ABS
control. The wheel brake generates the braking power still
necessary. In this condition, the brake force distribution
onto the axles is not changed in comparison to the end of
the ABS control. A new ABS entry is not possible with
unmodified marginal conditions (coefficient of friction,
braking demand).
5. Upon expiry of 4, the limit for the brake torque of the
regenerative brake continues to be slowly raised until the
maximum value predetermined by the generator is reached.
The wheel brake generates the braking power exceeding
the maximum value. In this condition, the brake force
distribution to the axles is gradually shifted back to the
driving axle. A new ABS entry is possible with unmodified
marginal conditions (coefficient of friction, braking demand).
6. Upon expiry of 5, the regenerative brake is fully
activated again. Thus, item 1 is resumed.
To avoid permanent recurring cycles, the number of runs
is limited as follows:
With a new ABS entry in state 5, a stop is made in 4
during the next run (no more transitions to 5). The limit for
the brake torque of the regenerative brake hence remains
on the value reached in 4.
With a new ABS entry in state 4, a stop is made in 3
during the next run (no more transition to 4). The regenerative brake stays deactivated.
With a new ABS entry in state 3, the number of runs is
not limited. There is no detrimental activation and deactivation of the regenerative brake.
Hence, the following runs are possible after an ABS
entry: 1-2-3-4-5-1, 1-2-3-4-5-2-3-4, 1-2-3-4-2-3, 1-2-3-45-2-3-4-2-3 and 1-2-3-2-3-2 When the cycle is stopped
as described, a timer is started. Upon expiry of a long period
of time (e.g. 30 seconds), or when the driver accelerates, or
no more brake torque is demanded, the cycle is released
and continued at the point where it was halted. The period
is chosen to be so long that one can assume that the driver
no longer notices the correlation as disturbing.
2.5. Control of Regenerative and Hydraulic Braking Force
Distribution

IMPROVEMENT OF DRIVABILITY AND FUEL ECONOMY WITH A HYBRID ANTISKID BRAKING SYSTEM 209
The ideal braking force distribution relation between the
front and rear axles of a conventional vehicle is

G b hg(a b) Fuf Gb Fuf

(23)
hg

G
h
Where, hg is the distance from gravity center of sprung
mass to roll center, Fuf and Fur are respectively the hydraFur

1
= -- ---2
g

+
4
+ -----------------------

------- + 2

ulic braking force of the front and rear axles.

In fact, a conventional vehicles braking force distribution ratio, , is found according to the more practical
relation:

F
RD
(24)
Fuf Fur Rf Df Rr Dr
Where Rf and Rr are respectively the front and rear brake
effective radius, Df and Dr are the front and rear brake
2

= -----------uf------- = -----------2--f------f---------2
+

wheel cylinder diameter respectively.

However, for a hybrid electric vehicle, the braking
torque distribution between the front and rear axles should
be changed to account for the regenerative braking torque.
It is assumed that the regenerative braking torque is added
to the front axles. Therefore, the hybrid electric vehicle
braking force distribution ratio, hyb, should be

Fuf Fmotor_reg
Fuf Fmotor_reg Fur
+

hyb= -----------------------------------------
+

(25)

According to the regulations of ECE R13, the distribution ratio of the hybrid electronic vehicle braking force
should be

b zh
L
(z

(26)

hyb --------------g
+

hyb

b zh
zL
) ( a zhg )
zL

+ 0.07 ) ( +
g)
----------------------------------------0.85

+ 0.07

hyb 1 ----------------------------------------0.85

Rf Df
Rf Df Rr Dr
Wherein Z is braking intensity (z =0.1~0.61).
2

hyb -----------2-----------------2
+

(27)
(28)
(29)

(V0, Vw0). The parameter a is the gradient of the curve.

So, for an almost adhesive wheel (0<<1.0), Padh(s) can be
described as

Padh(s ) G gI g s
w

(33)

1
= ---------------- -+

On the other hand, for the completely skidding wheel

( 0 ), the dynamics seem to be Pskid(s),
1.0

Pskid(s ) I s
w

(34)

11
- -= ---

The hydraulic system generates braking torque Fhydra_ABS

based on the hydraulic braking torque reference Fhydra (see
Figure 4). The total braking torque Fbrake_hyb is
(35)
Fbrake_hyb F*motor_reg Fhydra_ABS
Normally, the ABS controller does nothing, therefore,
(36)
Fhydra_ABS Fhydra
Wherein denotes the uncertainty factor such as pad
secular variation.
In this case, motor torque should be
*
(37)
Fmotor_reg Fmotor_reg
Wherein both Fmotor_reg and Fhydra are required by the
braking force distribution strategy.
With a new ABS entry in state IV, V and VI the feedback controller should vary F*motor_reg actively and dynamically. This feedback controller can be described with Pn(s)
and Qmotor_FB(s ) ,
=

Pn(s) (G gI g)s
w
Qmotor FB(s ) Gg s s
= ----------------------+

1
- ------------= --+1

s
s

w +1
= ---------------------------------------- -------------+
1 0)]
a +1

Wherein

IV

(30)

a= ---w------w---0 --------------------------------aN G( 1 0) + Iwg

(31)

w= ---------w---0

(32)

GV
aNg

Wherein 0 is a slip ratio at the same operational point

(39)

With these feedback controllers and , which is the time

constant of controller, the transfer functions from Fhydra_ABS
to Fhydra_ABS and from Fhydra_ABS to Vw can be changed
according to their Bode diagrams, which can be calculated
with Padh( s) and Pskid( s) , respectively (Shin-ichiro and

2.6. Regenerative Braking Controller Design

The transfer function from motor regenerative braking
force Fmotor_reg to the wheel velocity Vw is

P(s) [ I g Gg(
w

(38)

Figure 4. Block diagram of proposed controller.

210

J. L. ZHANG, CH. L. YIN and J. W. ZHANG

Yoichi, 2000).
To prevent the transmission of hydraulic braking torque
being blocked with motor torque, the feed forward controller motor_FF is designed to be:

hi =

G F +F
Cmotor_FF( s )= ------------------2G + Iwg hydra motor_reg

Tmotor_cont is the continuous torque of motor.

(40)

when the braking mode changes from the regenerative

braking mode to the hydraulic mode.
Because the hydraulic pressure for braking torque is
not yet sufficient, the regenerative braking torque is
diminished gradually. The motor load command at time ,
MC_cmd( t ) , should be
min(Tmotor_reg_max( t), Tmotor_reg(t 1 ) )
- (41)
MC_cmd( t )= -----------------------------------------------------------------------------------------

Where

Wherein is attenuation gradient of regenerative braking torque, Tmotor_reg_max(t) is the current available maximal
regenerative braking torque of the motor at time and
Tmotor_reg( t 1 ) is the regenerative braking torque of the
motor at time 1.
t

2.7. Analysis of the Factors Influencing Regenerative Braking Torque

The factors that influence regenerative braking torque are
mainly battery state of charge (SOC), motor speed and
motor temperature.
With respect to battery state of charge, SOC is defined as
battery SOC impact factor of regenerative braking torque.
h1 and
h2 are its threshold values. It is defined as:
K

SOC

SOC

1
SOC SOCh1
KSOC= --------------------------------- SOCh2 SOCh1
0

SOC < SOCh1

(42)

SOCh1 < SOC < SOCh2

SOC > SOCh2

As for motor speed, m, denotes the motor speed

impact factor of regenerative braking torque. m1 and m2
are its threshold values. It is defined as:
K

1
m m1
K= ------------------- m2 m1
0

m > m2
m1 < m < m2

(43)

m < m1

Concerning motor temperature, the heat index, i, can be

used to represent the fact that the more the output torque of
the motor exceeds the continuous torque curve, the greater
the heat index will be until it reaches 1. The more the
output torque is below the continuous torque curve, the
smaller the heat index will be until it reaches 0. It is defined
as:
h

Tmotor_reg
- 1 h dt
y = -----------------------------Tmotor_cont( m ) c

(44)

hc

(45)

is the temperature rise constant of motor and

So, the available maximum regenerative braking torque

of the motor, Tmotor_reg_max , should be
Tmotor_reg_max

Tmotor_reg_max( t )

0 y<0
1 y>1
y other

=max

hi Tmotor_cont

+ 1 hi
(

) T motor_peak

U Im_max
, -----------------m

(46)

Wherein Tmotor_peak is the motor peak torque. Im_max and

are the maximum current and voltage respectively.

2.8. Fuzzy Antiskid Braking System

In this paper, the hybrid antiskid braking system is a multilayered control architecture with a slave controller (the fuzzy
antiskid braking controller) as the lower level modules.
The slip ratio, , is one of the output parameters in the
whole vehicle model. Usually, slip ratio is used to
evaluate the slip. In the fuzzy logic controller, a twodimensional rule table is created based on the error, ,
between the desired slip ratio and actual signals, and on the
change in the error, . The controller receives the signals
and as inputs and generates, as output, the hydraulic
change to control the hydraulic braking system. The slip
ratio error, is defined as e= d , where is the
vehicle slip ratio, and d is the desired vehicle slip ratio.
To drive the vehicle along the desired track while the slip
ratios cater to the desired vehicle slip ratio, the control
action must impose the behavior of the hydraulic braking
as a function of its actual state quantified by SLIP_
RATIO_ERROR and ERROR_CHANGE.
On the basis of experience and the requirements, the
membership functions of input and output were chosen as
Gaussian functions.
The linguistic variable SLIP_RATIO_ERROR has seven
linguistic terms, and ERROR_CHANGE has seven linguistic terms, leading to a maximum number of 49 terms that
can be achieved. The rules designed to control the vehicle
have the general format:
If (SLIP_RATIO_ERROR, ERROR_CHANGE) then
(HYDRAULIC_CHANGE).
The defuzzification method of the output variable
HYDRAULIC_CHANGE is centre-of-area.
The rules are listed in Table 1, where NB is Negative
Big, NM is Negative Medium, NS is Negative Small, ZO is
Zero, PS is Positive Small, PM is Positive Medium and PB
is Positive Big. The output of the fuzzy logic controller is
represented in Figure 5. The fuzzy controller structure is
represented in Figure 6. The role of each block is the
following:
The
converts the input values ( ,
) into linguistic terms of the input fuzzy variables
(SLIP_RATIO_ERROR, ERROR_CHANGE), with a
e

Fuzzification Interface

IMPROVEMENT OF DRIVABILITY AND FUEL ECONOMY WITH A HYBRID ANTISKID BRAKING SYSTEM 211

Table 1. Fuzzy controller rules table.

HYDRAULIC
_CHANGE NB
EG NB PB
NM PB
N
A
NS PM
H
C_ ZO PM
RO PS
PS
RR PM PS
E PB PB

SLIP_RATIO_ERROR
NM NS ZO PS PM
PB PM PS PM PM
PB PM PM PM PS
PM PS PS PS ZO
PS PS ZO NS NM
PS ZO NS NM NM
ZO NS NS NM NM
NS NM NM NB NB

Table 2. Parameters in the simulations.

PB
PM
PS
ZO
NB
NB
NB
NB

Vehicle weight
Wheel inertia
Wheel radius
Wheelbase
Wheel tread front/rear
Gear ratio
Max. of hydraulic braking torque
Max. of engine torque
Engine speed at max. torque
Max. of engine power
Engine speed at max. power
Max. of regenerative braking torque
Battery nominal capacity
Battery total voltage
1st order delay in motor torque response

1080[kg]
1.1858[kgm2]
0.304[m]
2.620[m]
1.465/1.455[m]
3.417
405[Nm]
142[Nm]
4000[rpm]
78[kW]
5750[rpm]
250[Nm]
8[Ah]
288[V]
0.001[s]

Figure 5. Output of fuzzy logic controller.

Figure 6. Fuzzy controller.

Figure 7. Simulation model for the control system.

corresponding certainty value,

The
stores the data that defines the input
and the output fuzzy sets, as well as the fuzzy rules that
describe the control strategy,
The
block applies the fuzzy rules to the
input fuzzy variables to obtain the output values
(HYDRAULIC_CHANGE),
The
achieves output signals
based on the output fuzzy sets obtained as the result of
fuzzy reasoning.

can perform the simulation calculation of various control

methods. The main program mainly includes the motor
controller module, hydraulic fuzzy controller module,
hydraulic and actuation system module, and the non-linear
whole vehicle module.
With the data file of a tested vehicle loaded, simulation
tests can be carried out for the whole braking process of the
vehicle. A monitor, i.e., Scope, is designed to graphically
depict the output results for each subsystem in real-time.
The initial velocity before braking is 90 Km/h. The
friction coefficient of the road is 0.5. Typical simulation
results are presented in this paper. The calculation results
are shown in Figure 8. The performance variables including vehicle speed and wheel spin speeds are presented in
Figure 8(a), hydraulic braking and motor regenerative
braking torque in Figure 8(b), hydraulic braking pressure in
Figure 8(c), wheel slip ratio in Figure 8(d) and regenerative
energy and battery SOC changing in Figure 8(e).
From Figure 8(a), we see that the control logic can
prevent wheel locking. The slip ratio is maintained near the
desired value, and its oscillation can be suppressed (see
Figure 8(d)). Because of the significant reduction of brak-

Knowledge Base

Decision Logic

Defuzzification

Interface

3. SIMULATION RESULTS
Simulations are carried out to confirm the effectiveness of
the proposed controller. Table 2 shows the parameters in
the simulations.
Figure 7 shows the closed-loop simulation model built
by CRUISE software developed by AVL, which is the
superior software for the simulation research on vehicle
dynamic, economic, exhaust and braking performances.
The control logics and strategies of the vehicle components are built using Matlab/Simulink (Natick, 1990), which

212

J. L. ZHANG, CH. L. YIN and J. W. ZHANG

Figure 9. Hybrid components.

less frequently than that of front wheels.
When braking, for the sake of avoiding wheel locking,
there are some adjustments of hydraulic braking torque and
motor regenerative braking torque (see Figure 8(b)). The
cooperation between regenerative and hydraulic braking is
good. During the braking course, state of charge (SOC) of
battery changes from 45% to 45.5% (see Figure 8(e)). The
increase of SOC means that vehicle fuel economy can be
improved.
In summary, the simulation results show that the proposed control strategy and method are feasible. Its robustness is proven by the simulation results. The best compromise between braking stopping distance and stability
performance for the vehicle can be obtained.

4. EXPERIMENTAL RESULTS
This proposed control system is examined experimentally
with one HEV (Figure 9).
Some typical test results are presented in this paper. The
vehicle braking field test results for vehicle velocity and
wheel speed on a low adhesion coefficient road are shown
in Figure 10(a), wheel slip ratio in Figure 10(b), hydraulic
braking pressure and motor regenerative braking torque in
Figure 10(c), motor regenerative braking current and battery
SOC in Figure 10(d).
Under the coordinated control of regenerative and antiskid braking (see Figure 10(c)), we can see that the control
logic can prevent wheel locking and maintain the slip ratio
near the desired value (see Figure 10(b)). As a result, the
final state of charge (SOC) value is higher (see Figure
10(d)). The kinetic energy received during this process can
be used to recharge the traction battery and be stored for
future use.
Figure 8. Simulation results on low adhesion coefficient
road with FC strategy.
ing friction coefficient, braking pressure decreases (see
Figure 8(c)). Because of the employed fixed slip ratio control, the braking pressure of the rear wheels is modulated

5. CONCLUSION
This paper put forward a novel control strategy for a regenerative braking and antiskid braking system for hybrid
electric vehicles. The simulation and experimental results
show that the proposed control strategy is robust and effective for improving vehicle drivability performance and fuel

IMPROVEMENT OF DRIVABILITY AND FUEL ECONOMY WITH A HYBRID ANTISKID BRAKING SYSTEM 213

controllers and combinations of fuzzy control and artificial

neural networks. In addition, control schemes that seek to
optimize as many sources as possible will be employed.
Research and development of these technologies must proceed together with the continuous collaboration between
automotive engineers and control engineers. The practical
ends of intelligent control combined with a deep understanding of mechatronics will be increasingly important in
future vehicle engineering endeavors.

ACKNOWLEDGEMENTThe research presented is financi-

ally supported by National Development and Reform Commission (NDRC), Peoples Republic of China under the project
contract (No. 2005934).

REFERENCES
Baumann, B., Washington, G., Glenn, B. and Rizzoni, G.
(2000). Mechatronic design and control of hybrid electric
vehicles. IEEE/ASME Trans. Mechatronics, , 5872.
Benbouzid, M. E. H., Diallo, D., Zeraoulia, M. and Zidani,
F. (2006). Active fault-tolerant control of induction motor
drives in ev and hev against sensor failures using a fuzzy
decision system. Int. J. Automotive Technology , , 729
739.
David, E., Mohammed, F., Robert, K. and Bert, H. (1994).
Fuzzy anti-lock brake system. Intel Corporation, Automotive Operation, Arizona.
Jochen, F. and Gregor, S. (2006). Co-ordination Method for
a Regenerative and Anti-skid Braking System. United
States Patent Application 007152934 B2. 2006-12-26.
Jung, H. G., Hwang, J. Y., Yoon, P. J. and Kim, J. H. (2007).
Resistance estimation of a PWM-driven solenoid. Int. J.
Automotive Technology , , 249258.
Mokhiamar, O. and Abe, M. (2006). How the four wheels
should share forces in an optimum cooperative chassis
control. Control Engineering Practice, , 295304.
Natick, M. (1990). Matlab Users Guide. Mathworks, Inc.
Oh, K., Kim, D., Kim, T., Kim, C. and Kim, H. (2005).
Efficiency measurement and energy analysis for a HEV
bench tester and development of performance simulator.
Int. J. Automotive Technology , , 537544.
Pacejka, H. B. (2002). Tyre and Vehicle Dynamics.
Butterworth-Heinemann Publications. Oxford.
Sakai, H. (1990). Study on cornering properties of tire and
vehicle. Tire Science and Technology , , 136169.
Shin-ichiro, S. and Yoichi, H. (2000). Advanced vehicle
motion control of electric vehicle based on the fast motor
torque response. Proc. 5th AVEC, 729736.
Yu, F., Feng, J. Z. and Li, J. (2002). A fuzzy logic controller design for vehicle abs with a on-line optimized
target wheel slip ratio. Int. J. Automotive Technology ,
, 165170.
Zhao, Y. S., Chen, L. P., Zhang, Y. Q. and Yang, J. (2007).
Enhanced fuzzy sliding mode controller for launch control of AMT vehicle using a brushless DC motor drive.
Int. J. Automotive Technology , , 383394.
5

7 6

8 2

14

6 5

18

Figure 10. Experimental results on low adhesion coefficient road.

economy. It can assure braking safety and capture more
regenerative energy. Hydraulic braking torque and regenerative braking torque can be coordinated with each other
under this control strategy and method.
Future work in this field will include adaptive fuzzy

8 3