sustainability

Article
Interval Type-2 Fuzzy Logic Anti-Lock Braking Control for
Electric Vehicles under Complex Road Conditions
Linfeng Lv 1 , Juncheng Wang 1, * and Jiangqi Long 2

1 Faculty of Mechanical Engineering and Automation, Zhejiang Sci-Tech University, Hangzhou 310018, China;
ysulvlinfeng@163.com
2 School of Mechanical and Electrical Engineering, Wenzhou University, Wenzhou 325035, China;
longjiangqi@163.com
* Correspondence: wangjc90@163.com; Tel.: +86-157-5101-0621

Abstract: To simultaneously track the ideal slip rate and realize ideal energy recovery efficiency
under different complex road conditions, an electro-hydraulic compound anti-lock braking system
based on interval type-2 fuzzy logic control strategy and its corresponding braking torque allocation
strategy have been developed for electric vehicles. The proposed interval type-2 fuzzy logic controller
aims to calculate the ideal total braking torque by four steps, namely, fuzzification, fuzzy inference,
type reduction, and defuzzification. The slip rate error and the change rate of slip rate error are
utilized as inputs in the fuzzification, and then, the membership degree interval of fuzzy variables
determined by the upper and lower membership functions is used to calculate the activation degree
interval of different fuzzy rules in the fuzzy inference process, which enhances the anti-interference



ability to external uncertainties and internal uncertainties. The braking torque allocation strategy is

 proposed to maintain the maximum energy recovery efficiency on the premise of safe braking. The
Citation: Lv, L.; Wang, J.; Long, J. software of MATLAB/Simulink is applied to simulate the process of anti-lock braking control under
Interval Type-2 Fuzzy Logic two complex road conditions. Simulation results corroborate the proposed interval type-2 fuzzy logic
Anti-Lock Braking Control for anti-lock braking control system can not only obtain better slip rate control effect and outstanding
Electric Vehicles under Complex robustness but also achieve ideal regenerative braking energy recovery efficiency under both joint-µ
Road Conditions. Sustainability 2021, and split-µ road surfaces.
13, 11531. https://doi.org/10.3390/
su132011531
Keywords: electric vehicles; anti-lock braking system (ABS); interval type-2 fuzzy logic; regenerative
braking; slip rate control; energy recovery
Academic Editors: Jiankun Peng,
Fengyan Yi, Dawei Pi and Yue Wang

Received: 9 September 2021

Accepted: 14 October 2021
1. Introduction
Published: 19 October 2021 In the face of the problems of environmental pollution and energy shortage, electric
vehicles (EVs) have become the main development direction of the automotive industry [1].
Publisher’s Note: MDPI stays neutral Taking advantage of both regenerative braking subsystem in braking energy recovery
with regard to jurisdictional claims in and traditional hydraulic braking subsystem in high power density, the electro-hydraulic
published maps and institutional affil- compound braking system has great potential to make a trade-off between energy saving
iations. and braking control effect for EVs [2]. With the increase of both driving velocity and
driving condition complexity of EVs, the anti-lock braking system (ABS) is regarded as one
of the necessary auxiliary systems of electric vehicles to decline the rate of traffic accidents.
A cooperative control algorithm for electric vehicles was proposed to improve the ABS
Copyright: © 2021 by the authors. working performance [3]; additionally, a nonlinear model predictive control was used on
Licensee MDPI, Basel, Switzerland. ABS and tested on a high-end vehicle simulator to verify the performance and robustness
This article is an open access article of the ABS controller [4]. Compared with traditional vehicles, the EVs can implement
distributed under the terms and regenerative braking technology to enhance energy utilization efficiency by recovering
conditions of the Creative Commons kinetic energy during the braking process [5]. Thus, the anti-lock braking function of
Attribution (CC BY) license (https:// EVs can be done by the electro-hydraulic compound braking system to make a trade-off
creativecommons.org/licenses/by/ between the energy saving and anti-lock braking control effects.
4.0/).

Sustainability 2021, 13, 11531. https://doi.org/10.3390/su132011531 https://www.mdpi.com/journal/sustainability

Sustainability 2021, 13, 11531 2 of 23

The ABS control strategy can be regarded as one of the main key technologies for EVs
to realize the anti-lock braking function, which is used to calculate the demand braking
torque of every wheel [6]. At present, some traditional control algorithms have been widely
applied in designing the vehicle ABS controller, which is mainly including logic threshold
control, proportion integral differential (PID) control, and sliding mode control. Chi-
ang et al. [7] proposed a logic threshold ABS control system to realize the anti-lock control
function by constraining the motor reference torque with an allowable value. Yang et al. [8]
designed a logic threshold control with phase plane theory to analyze the relation between
slip rate and braking torque, and then, the composition rule of expected braking torque
was studied to improve the anti-lock braking control effect. However, by analyzing these
ABS control strategies on the basis of the logic threshold control method, it can be seen that
the rule database for various road conditions is difficult to be established. Feng proposed
the discrete fuzzy adaptive PID to accurately track the ideal wheel slip rate [9]. Moreover,
to enhance the robustness of a control system regardless of system parameter variations,
sliding mode control is still widely applied in ABS control. Sun et al. [10] investigated a
sliding mode wheel slip rate control to yield anti-lock control of wheels with an adaptive
sliding surface, and Rahul et al. [11] proposed a multiple surface sliding controller to
maintain the optimal slip rate in unknown road surface conditions. However, note that
the common sliding mode ABS controller is always robust but not optimal. In summary,
the control effects of traditional control algorithms always depend on the accuracy of
mathematic modeling. Due to the massive nonlinear, time-varying and lagging influencing
factors in the anti-lock braking control process, the mathematical dynamics model of ABS
is difficult to be accurately described, especially for EVs.
With the rapid development of intelligent technology, intelligent control algorithms
have great advantages, which mainly include fuzzy control [12], neural network [13], and
genetic algorithm [14]. Due to a large number of nonlinear, time-varying, and hysteretic
factors in the process of vehicle ABS control, the ABS control model is difficult to be
accurately described. Therefore, fuzzy control, which does not depend on the precise
mathematical model of the controlled object, has been widely studied by experts and
scholars. Fargione et al. [15] proposed a fuzzy control strategy integrated optimization of
genetic algorithm to realize the anti-lock braking function of the electro-hydraulic braking
system. Andrei et al. [16] improved the vehicle braking stability and regenerated the
maximum possible amount of energy by designing a fuzzy control algorithm on the basis of
road recognition. Mokarram et al. [17] studied a fuzzy logic controller in 0.35 µm standard
complementary metal oxide semiconductor (CMOS) process and used adaptive neural-
fuzzy inference systems of software to define the parameters of the fuzzy logic controller;
the simulation results show the controller have a high speed of calculation and low power
consumption in ABS. However, the proposed controller lacks adaptive ability because the
fuzzy logic parameters are invariable. In summary, the shape of the membership function
and the corresponding membership degree of each point in the domain for the fuzzy logic
control algorithm mentioned above are determined, so it can be collectively referred to
as ‘type-1 fuzzy logic control’. However, the shape of the membership function and the
membership degree corresponding to each point in the domain are single and invariable
in the type-1 fuzzy logic controller. Moreover, in the process of EVs anti-lock braking
control, the information of different road adhesion coefficient and optimal slip rate has
strong uncertainty, and the type-1 fuzzy logic control is lack of adaption for environmental
variation with more uncertain information. Therefore, the type-1 fuzzy logic control has
unsatisfactory performance in tracking optimal slip rate and energy recovery when road
surface abruptly changed or the EVs wheels braking on different road surface respectively.
On the basis of the traditional fuzzy set, the type-2 fuzzy set has carried on the
expanded dimension processing. A single fuzzy variable is described by two different levels
of membership function, which can simultaneously mode both intra-personal uncertainty
and inter-personal uncertainty [18,19]. Hence, in many applications, such as system
controlling, decision making, and machine learning, the type-2 fuzzy control algorithm
Sustainability 2021, 13, 11531 3 of 23

have been demonstrated better performances compared with the traditional type-1 fuzzy
control. Claudia I et al. [20] proposed a generalized type-2 fuzzy logic system with the
limitation of complexity by the theory of alpha-planes. Zhang [21] used trapezoidal interval
type-2 fuzzy sets to investigate the multiple attribute group decision-making problems.
Gaxiola [22] et al. used an improved type-2 inference system to estimate the type-2
fuzzy weights of backpropagation neural network, and the simulation results illustrate the
advantages of the bio-inspired methods optimizing type-2 fuzzy systems. Sanchez et al. [23]
used a mobile robot in conjunction with three types of external perturbations to contrast
the control performance of generalized type-2 fuzzy systems and interval type-2 fuzzy
systems and type-1 fuzzy systems; the results show the type-2 fuzzy logic control has better
anti-interference ability than type-1 fuzzy logic control. It is noted that there has been no
research about the ABS control method based on type-2 fuzzy logic algorithm until now.
The possible reasons are that the computing process is complex and time-consuming, which
is unfit for the time-varying ABS model, and the fuzzy rules are difficult to be formulated
without sufficient experience.
This present study aims to fulfill the excellent optimal slip rate tracking effect under
complex road conditions by introducing an electro-hydraulic compound anti-lock braking
control system. The proposed anti-lock braking control system utilizes an interval type-2
fuzzy logic algorithm to calculate the expected braking force tracking ideal slip rate, which
differs from conventional type-1 fuzzy logic control. The membership degrees of fuzzy
variables in different fuzzy sets of conventional type-1 fuzzy logic are constant, which
reduce the anti-interference ability of the controller, resulting in unsatisfactory performance
of slip rate control when the working conditions are variable. The proposed control
algorithm utilizes upper and lower membership functions to describe the membership
degree of fuzzy variables so that the anti-interference ability and adaptation of the controller
can be enhanced when external conditions are changing.
This research considers the following contributions:
(1) The structure composition and operating principle of the proposed interval type-2
fuzzy logic electro-hydraulic compound anti-lock braking system is given out in
detail, and the allocation strategy is designed by considering the balance between the
energy recovery efficiency and braking safety.
(2) Considering the uncertain road conditions of anti-lock braking control, the single
fuzzy variable is described by membership function of two different levels by using
the membership function expansion method and set the secondary membership de-
gree of fuzzy variable to a constant value of 1 to enhance the ability of anti-interference
for fuzzy control under massive uncertainty information during the braking process,
and Karnik–Mendel (KM) algorithm fuzzy type reduction method is adopted to solve
the complex calculation problem of generalized type-2 fuzzy reasoning.
The present paper is organized as follows: Section 2 describes the modeling of a total
system. Section 3 depicts the design process of the interval type-2 fuzzy logic anti-lock
braking control system. Section 4 provides the simulation results. Section 5 presents the
conclusion drawn from the study.

2. System Model
2.1. Dynamic Model of Automobile Brake System
Establish the vehicle coordinate system consolidated to the center of mass, with the
x-axis pointing forward parallel to the ground, the y-axis pointing forward parallel to the
driver’s left, and the z-axis pointing upward through the center of mass.
Ignoring the dynamic effects of suspension, the vehicle only moves parallel to the
ground, and ignoring the changes in tire characteristics caused by load changes of left and
right tires and the effect of tire righting torque, the vehicle is simplified to the dynamic
model [24,25] shown in Figure 1.
Sustainability 2021, 13, x FOR PEER REVIEW 4 of 24

Sustainability 2021, 13, 11531 4 of 23

and right tires and the effect of tire righting torque, the vehicle is simplified to the dy-
namic model [24,25] shown in Figure 1.

Fy4 Fx4 Fy2 Fx2

δ4 δ2 ω
vy
γ v Tb Td
tr vx tf

Fy3 Fy1 Fx1

Fx3
Fx
δ3 δ1
Tf
lb la Fz

(a) (b)
Figure 1. Vehicle
Figure dynamic
1. Vehicle model:
dynamic (a) (a)
model: dynamic model
dynamic of electric
model vehicle
of electric and
vehicle (b)(b)
and dynamic model
dynamic of tire.
model of tire.

TheThe
equation of motion
equation of the
of motion electric
of the vehicle
electric is expressed
vehicle as as
is expressed follows:
follows:


m  vx  v y     Fm
 4
4

 Xi vF
. cos .

x D− vy γ = ∑ FXi − FD cos β
 i 1


 i =1


m  v y  vx     FYimvFyD+
 4
. .
4
sinv xγ = ∑ FYi − FD sin β





 2 i 1 i =1



I d γ2
z dtd 2 = l ( F + F ) − l ( F + F Y4 ) +
t t
tf 2f ( FX1 − FX2 )tr+ 2r ( FX3 − FX4 ) (1)
       X 3  FX 4 
a Y1 Y2 b Y3
  I  l F  F  . l F  F  F  F  F (1)
z a Y1
JWi ω i = Tdi − Tbi − 2Fxi Rωi − Tfi 2
Y2 b Y3 Y4 X1 X2
 2
 dt



     ωi  T

J T T F xi R= Fx − Ff i cos δi − Fyi sin δi


Xi
 i
W i di bi
 fi i


 
 
  F  F  F cosF 

= F 
− F sin δi + Fyi cos δi

Yi F sin
x f
 Xi xi fi i yi i i i

where m isthe Yi
mass xi
 F  F  F sin   F cos 
fi
of the  i
electric yi
vehicle; i
. .
vx , vy , v x , and vy represent the velocity and
acceleration of electric vehicle along the x- and y-axis, respectively. β is the angle between
where m is the mass of the electric vehicle; vx, vy, v x , and v y represent the velocity . and
the air resistance and the driving direction. FD illustrates the air resistance. γ, γ, and Iz are
acceleration
the vehicle’s of electric
yaw angle, vehicle yawalong angular thevelocity,
x- and y-axis, and moment respectively.
of inertia β isaround
the angle be-
the z-axis,
tween the air resistance
respectively. la and lb are andthe the driving between
distances direction. theFDmassillustrates
center of thetheairvehicle
resistance. γ, front
and the γ,
andand rearthe
Iz are axles, respectively.
vehicle’s yaw tf and
angle, yaw tr indicate,
angular respectively,
velocity, and the front and
moment of rear wheelbases.
inertia around
.
theJW , ωi , and
z-axis,
i
ω i stand for
respectively. la and thelmoment
b are the of inertia, angular
distances betweenvelocity,
the massand angular
center of the acceleration
vehicle
of the wheel, respectively.
and the front and rear axles, respectively. T di , T bi , and T ftif and tr indicate, respectively, the front torque,
are the wheel driving torque, braking and
and
rear rolling resistance
wheelbases. J W , ωtorque,
i
, and respectively.
 i
stand for R
the ω i
is the
moment rolling
of radius
inertia, of
angularthe wheel.
velocity,Fx i
and δi
and
i
are the longitudinal force, lateral force, and steering angle of the wheel, respectively. The
angular acceleration of the wheel, respectively. Td , Tb , and Tf are the wheel driving
subscript i are 1, 2, 3, and 4, indicating the front right wheel, front left wheel, rear right
i i i

torque,
wheel, braking
and rear torque,
left wheel,and rolling respectively. resistance torque, respectively. Rω is the rolling ra- i

dius of the wheel. Fx , Fy , and  i are the longitudinal force, lateral force, and steering
2.2. Tire Model i i

angle of the wheel, respectively. The subscript i are 1, 2, 3, and 4, indicating the front
The Magic Formula tire model [26] is used to describe the dynamic behavior of vehicle.
right wheel, front left wheel, rear right wheel, and rear left wheel, respectively.
Y = D sin{Carctan{ B(λ + SH ) − E[ B(λ + SH ) − arctanB(λ + SH )]}} + SV (2)
2.2. Tire Model
where Y is theFormula
The Magic output variable,
tire model namely
[26] islongitudinal
used to describe force;the
λ isdynamic
the slip rate of theofelectric
behavior ve-
wheel; B, C, and E represent the stiffness, shape, and curvature coefficients, respectively;
hicle.
D is the peak value; SH and SV stand for the horizontal and vertical biases, respectively.
Y rate
The slip D sinofCthe
arctan
electric 
B  wheel
 SH   
 B    as
is Edefined   arctan B    SH   SV
SH follows: (2) 
v x − Rωi ωi
λi = (3)
vx
Sustainability 2021, 13, 11531 5 of 23

2.3. Hydraulic Braking System Model

The typical dynamic model of ABS hydraulic system is described as follows [27]:

dPw 1 √
1 √ 
0

= Pm − Pw u1 t − τvp − Pm − Pr u 2 t − τvp (4)
dt Ce Re Ce Re0

where Pm , Pw , and Pr are the pressure of the main cylinder, wheel cylinder, and low-
pressure accumulator, respectively; Ce is the equivalent liquid capacity characteristic
coefficient of the pipeline and wheel cylinder; Re and Re0 are the equivalent liquid resistance
characteristic coefficient of the pipeline and wheel cylinder when the pressure is increased
and reduced, respectively; τvp and τvp 0 are the transmission lag time of solenoid valve and

pipeline during pressurization and decompression, respectively; u1 and u2 are the control
command signal of solenoid valve:

 u1 = 1, u2 = 0 system presure increasing
u = 0, u2 = 1 system presure decreasing (5)
 1
u1 = 0, u2 = 0 system presure maintaining

Considering the influences of oil pressure, friction coefficient, temperature, and other
factors, the first-order inertia link is used to represent the time delay in the process of
applying brake pressure to the actual braking torque output:

1
P(s) = P0 (s) (6)
τs + 1
where P(s) and P0 (s) are the actual and target oil pressure of the brake, respectively; τ is a
constant reflecting the dynamic characteristics of the brake.
The hydraulic braking torque Th produced by the brake can be written as follows:

Th = 2Pw Af µb ηRb (7)

where Af indicates the area of the brake wheel cylinder piston. µb represents the friction
coefficient; η is wheel cylinder efficiency, and Rb stands for the effective radius of friction.

2.4. Regenerative Braking Dynamics Model

In this research, the front and rear wheels of the electric vehicle are equipped with
an in-wheel motor which can drive/brake independently of each other. The type of the
in-wheel motor is set as the permanent magnet synchronous motor (PMSM), and the
mathematical model of the motor is expressed as follows:
 . pω i L

 id = uLd − RLΩ id + dL q q
 . d d d
uq RΩ iq pωd id Ld ψf pωd
i q = − − − (8)
 Lq Lq Lq Lq
 
 T = 1.5p ψ i + L − L i i 

r f q d q d q

where ud and uq are the voltage of d and q axes, respectively; id and iq indicate the current of
d- and q-axis, respectively; Ld and Lq represent the inductor of d- and q-axis, respectively; p
is the number of pole-pairs; ψf illustrates the motor magnetic chain; ω d is the rotor angular
velocity of the motor, and Tr is the braking torque of the motor.
The in-wheel motor is adopted with a round rotor structure, and the Ld equals to Lq ,
which means the Tr can be simplified as follows:

Tr = 1.5pψf iq (9)

Thus, the braking torque of the motor can be gained by controlling the iq of q-axis. The
tracing torque control theory of PMSM’s can be described in Figure 2. The θ is the rotation
angle of rotor, the iqi and idi are the expected current of q- and d-axis, respectively; the iai ,
 which means the Tr can be simplified as follows:
Tr  1.5 p f iq

Sustainability 2021, 13, 11531 Thus, the braking torque of the motor can be gained by controlling6 ofthe 23 iq of q-
The tracing torque control theory of PMSM’s can be described in Figure 2. The θ i
rotation angle of rotor, the iqi and idi are the expected current of q- and d-axis, respec
ily; the iiciai,are
bi , and ibi,the
and ici are the
expected expected
currents currents
of phase ofc,phase
a, b, and a, b, and
respectively; ia , ic, respectively;
b , nd ic are the ia, ib,
are the
actual actualofcurrents
currents ofand
phase a, b, phase a, b, and c, respectively.
c, respectively.

1 iai
1.5 p f
Three-phase
idi ibi PWM
inverter
θ ici
ia
ib
ic

PWSM

Figure2.2.Scheme
Figure of PMSM’s
Scheme torque
of PMSM’s tracking
torque control.control.
tracking

The coordinate transformation formula of three-phase expected current iai , ibi , and ici
The coordinate transformation formula of three-phase expected current iai, ibi, a
is described as follows:
is described as follows:
− sin θ

   
iai cos θ

 ibi  =  cos θ − 2 π idi
2
3
− sin θ − 3 π

  (10)
iqi 
ici cos θ + 23 πcos−
 sin θ + 2
3  sin 
π
iai   
Input the respective different 
 values between  2  , ibi , and ici and ia ,ibi,dand
iai 2 i ic into the
i   cos       sin         devices of
hysteresis current control unit to get
bi  the control
 the switch
3   3    qi 
switch signals of six switchingi
ici  control
the three-phase inverter, and then on and off. Finally, torque tracking
  
control of the PMSM motor is realized.cos   2    sin    2   
   
  3   3 
3. Design of Interval Type-2 Fuzzy Logic Anti-Lock Braking Control System
Input the
3.1. Overview respective
of Interval Type-2 different
Fuzzy Logicvalues
Control between
Strategy iai, ibi, and ici and ia, ib, and ic int
hysteresis current
Type-2 fuzzy control
set A unit to of
e is an extension gettype-1
the control
fuzzy set,switch
which hassignals
largerofmembership
six switching de
of the three-phase inverter, and then control the switch on and off. Finally, torque t
space, better anti-interference ability, and accurate ability of fuzzy logic to approximate
unknown function
ing control of thewhen
PMSM dealing withishighly
motor uncertain information. The type-2 fuzzy set
realized.
Ae can be described as follows [28]:


Ae = ( x, u), µ e ( x, u) | x ∈ X, u ∈ Jx ⊆ [0, 1] (11)
A

where x and u are respectively the first and second variables of the type-2 fuzzy set; Jx and
µ Ae ( x, u) are respectively the primary and secondary membership degree.
The membership function of the type-2 fuzzy set is composed of upper membership
function (UMF) and lower membership function (LMF), and the corresponding relation-
ships are shown in Figure 3.
The interval type-2 fuzzy set is a special case of type-2 fuzzy set where the secondary
membership degree of variable x is identically equal to 1 and can be described as follows:

e = {(( x, u), 1)| x ∈ X, u ∈ Jx ⊆ [0, 1] }

A (12)
 where x and u are respectively the first and second variables of the type-2 fuzzy set; Jx
and A  x, u  are respectively the primary and secondary membership degree.
The membership function of the type-2 fuzzy set is composed of upper membership
Sustainability 2021, 13, 11531
function (UMF) and lower membership function (LMF), and the corresponding 7 of 23
rela-
tionships are shown in Figure 3.

 A  x, u 

 A  x1 , u 

1 u
x1
J x1
x
LMF A   UMF A  
Figure 3. The elements of a type-2 fuzzy set.
Figure 3. The elements of a type-2 fuzzy set.
Since the secondary membership degree of fuzzy variables in the interval type-2 fuzzy
setThe interval
is always equaltype-2 fuzzy
to 1, the set isofa massive
problem special calculation
case of type-2 in thefuzzy setofwhere
process type-2the second-
fuzzy
inference can be avoided, which means the results of fuzzy inference
ary membership degree of variable x is identically equal to 1 and can be described as can be obtained
quickly. At the same time, the description of the uncertainty between individuals in the
follows:
uncertainty domain composed of the upper and lower membership functions of fuzzy

variables enhances the abilityAof fuzzy 
 x, ulogic
 ,1 xtomodel
X , u  the 
J x anti-lock
0,1 braking system and (12)
the adaptability of the fuzzy logic to different external disturbances such as the abrupt
change of road adhesion coefficient and the optimal slip rate, and the interval type-2 fuzzy
Since the secondary membership degree of fuzzy variables in the interval type-2
logic has the advantage of the uncertainty processing ability to generalize type-2 fuzzy
fuzzy set is always equal to 1, the problem of massive calculation in the process of type-2
logic. Therefore, for the characteristics of complex, dynamic, nonlinear, time-varying, and
fuzzy inference
parameter can beofavoided,
uncertainty which
the anti-lock processmeans the results
in emergency of fuzzy
braking, inference
interval can be ob-
type 2 fuzzy
tained
logic quickly.
has betterAt the same
anti-lock time,control
braking the description
potential. of the uncertainty between individuals
in the uncertainty domain composed of the upper and lower membership functions of
3.2. Design
fuzzy variablesof Interval Type-2
enhances theFuzzy Logic
ability ofAnti-Lock Braking
fuzzy logic Controlthe
to model System
anti-lock braking system
and theThe working principle
adaptability of an electro-hydraulic
of the fuzzy logic to different compound
external anti-lock braking
disturbances system
such as the ab-
is to track the ideal slip
rupt change of road adhesion rate λ d coefficient and the optimal slip rate, and theand
by controlling the longitudinal slip rate of the wheel interval
make the fluctuation near the
type-2 fuzzy logic has the advantage value of of
thethe
ideal slip rate. The
uncertainty working ability
processing principletoofgeneralize
the
designed interval type-2 fuzzy logic electro-hydraulic compound anti-lock braking control
type-2 fuzzy logic. Therefore, for the characteristics of complex, dynamic, nonlinear,
system is shown in Figure 4.
time-varying,
As shown and in parameter
Figure 4, theuncertainty of the
interval type-2 anti-lock
fuzzy process in emergency
logic electro-hydraulic compound braking,
interval type
anti-lock 2 fuzzy
braking logic system
control has better anti-lock
is mainly braking
composed of control potential.
the interval type-2 fuzzy logic
controller, slip rate calculation unit, ideal slip rate identification unit, vehicle control unit
(VCU), regenerative braking system, hydraulic braking system, electric vehicle wheels, and
vehicle sensors.
The interval type-2 fuzzy logic controller takes the difference value e = λ − λd be-
tween the wheel actual slip rate and the ideal slip rate and its change rate as inputs and
then calculates the ideal anti-lock braking torque Tb_i through four steps, which include
fuzzification, fuzzy inference, type reduction, and defuzzification. Then, it outputs the
calculation results into VCU. VCU sends signals to the regenerative and hydraulic braking
systems to generate the regenerative and hydraulic braking torque on the wheels. The
wheel angular velocity and vehicle velocity are detected by the vehicle sensors, and vehicle
sensors output the result into the slip rate calculation unit and ideal slip rate identification
unit. The slip calculation unit and the ideal slip rate identification unit outputs the wheel
slip rate λi and the ideal slip rate λd to interval type-2 fuzzy logic controller, respectively.
 3.2. Design of Interval Type-2 Fuzzy Logic Anti-Lock Braking Control System
The working principle of an electro-hydraulic compound anti-lock braking system
is to track the ideal slip rate λd by controlling the longitudinal slip rate of the wheel and
make the fluctuation near the value of the ideal slip rate. The working principle of the
Sustainability 2021, 13, 11531 8 of 23
designed interval type-2 fuzzy logic electro-hydraulic compound anti-lock braking con-
trol system is shown in Figure 4.

λi +
λd
-
e e

Fuzzification

Slip rate
calculation unit Fuzzy
Fuzzy rules
v ωi inference
Road optimal slip
Type rate identification
reduction unit

Defuzzication Interval type-2

fuzzy logic
controller
Tb_i

Tr_i VCU Th_i

Regenerative Hydraulic
braking system braking system
Tr_r + + Th_r
Tb_r
Electeic
vehicle wheels

Vehicle
sensors
Figure4.4.Schematic
Figure Schematicdiagram
diagramofofinterval
intervaltype-2
type-2fuzzy
fuzzylogic
logicelectro-hydraulic
electro-hydrauliccompound
compoundanti-lock
anti-lock
brakingcontrol
braking controlsystem.
system.

As shown
The specificin Figure
design 4, the
steps interval type-2
of interval type-2fuzzy
fuzzylogic
logicelectric-hydraulic
electro-hydrauliccompound
compound
anti-lockbraking
anti-lock brakingcontroller
control system is mainly
are described composed
in detail of the interval type-2 fuzzy logic
as follows:
controller,
Step 1: slip rate calculation unit, ideal slip rate identification unit, vehicle control unit
fuzzification
(VCU),Theregenerative
Mamdani type is adopted
braking in the
system, proposed
hydraulic interval
braking type-2electric
system, fuzzy logic controller.
vehicle wheels,
The
anddifference value of the wheel slip rate and the ideal slip rate e = λ – λd and its change
vehicle sensors.
.
rate e The
are token
intervalas type-2
the input variables,
fuzzy and the ideal
logic controller takesanti-lock braking
the difference e = Tλb_i− is
torque
value λdthe
be-
output variable.
tween the wheel actual slip rate and the ideal slip rate and its change rate as inputs and
thenThe type-2 fuzzy
calculates set of
the ideal e is A
anti-lock = {NB, NS,
em braking ZE, PS,
torque Tb_iPB}; the subscript
through m values
four steps, whichare 1, 2,
include
3,fuzzification,
4, AND 5, whichfuzzystands in order
inference, forreduction,
type a subset of and
the type-2 fuzzy set. The
defuzzification. type-2
Then, it fuzzy set
outputs the
.
of e is B
en = {NB,
calculation NS,into
results ZE, VCU.
PS, PB};VCUthe subscript n values
sends signals areregenerative
to the 1, 2, 3, 4, andand
5 and respectively
hydraulic brak-
stand in order for a subset of the type-2 fuzzy set. NB, NS, ZE, PS, and PB represent the
negative big, negative small, zero, positive small, and positive big, respectively.
The type-2 fuzzy set of Tb_i is C ek = {SR, SM, MI, BI, BR}; the subscript k values are 1, 2,
3, 4, and 5 and respectively stand in order for a subset of the type-2 fuzzy set. SR, SM, MI,
BI, and BR indicate the smaller, small, middle, big, and bigger, respectively.
 spectively.
The type-2 fuzzy set of Tb_i is Ck = {SR, SM, MI, BI, BR}; the subscript k values are
1, 2, 3, 4, and 5 and respectively stand in order for a subset of the type-2 fuzzy set. SR,
Sustainability 2021, 13, 11531 SM, MI, BI, and BR indicate the smaller, small, middle, big, and bigger, respectively.
9 of 23
The upper and lower membership functions of inputs and output are gaussmf, ex-
pressed as follows:

The upper and lower membership functions of  x c 2

 inputs and output are gaussmf, ex-
pressed as follows: f  x ,  , c   e 2 2 (13)
2
− ( x −c)
where c determines the center position ) =the
f ( x, σ, cof 2σ2
e function; σ determines the width(13)
of the
curve of the function.
where c determines the center position of the function; σ determines the width of the curve
Figures
of the 5–7 illustrate the shape of the membership function of the input variables
function.
Figuresvariable.
and output 5–7 illustrate the shape of the membership function of the input variables and
output variable.

Sustainability 2021, 13, x FOR PEER REVIEW 10 of 24

Figure
Figure5.5.Membership
Membership function of
of e.e.

.
Figure
Figure 6.
6. Membership functionof
Membership function ofe. e .
Sustainability 2021, 13, 11531 10 of 23
Figure 6. Membership function of e .

Membershipfunction
Figure7.7.Membership
Figure Tb_i .
functionofofTb_i.

Step2:2:fuzzy
Step fuzzyinference
inference
The purpose of anti-lock braking
The purpose of anti-lock braking control
control isistotocontrol
controlthe thewheel
wheel slip rate
slip tracking
rate trackingthe
ideal slip rate. Therefore, when the different e between the wheel
the ideal slip rate. Therefore, when the different e between the wheel slip rate and the slip rate and the ideal
slip rate
ideal slip israte
negative and the
is negative absolute
and value isvalue
the absolute large, isthelarge,
ideal the
anti-lock
ideal braking
anti-locktorque Tb_i
braking
should be increased to increase the real braking torque. Therefore,
torque Tb_i should be increased to increase the real braking torque. Therefore, the wheel the wheel slip rate
increases
slip and approaches
rate increases the ideal
and approaches theslip rate.
ideal slipOn theOn
rate. contrary, when the
the contrary, whendifference e is
the differ-
positive and the absolute value is large, it is necessary to reduce the real
ence e is positive and the absolute value is large, it is necessary to reduce the real braking braking torque,
so thatso
torque, thethat
decreased wheel slip
the decreased rate
wheel is close
slip rate istoclose
the ideal
to theslip rate.
ideal Therefore,
slip the interval
rate. Therefore, the
type-2 fuzzy logic control rule is: if e is A em and e. is B en , then Tb_i is Cek .
interval type-2
Table fuzzythe
1 shows logic controlfuzzy
designed rule is: Am and e is Bn , then Tb_i is Ck .
if e is rules.
control
Table 1 shows the designed fuzzy control rules.
Table 1. Fuzzy control rules.
.
e
T b_i
NB NS ZE PS PM
NB BR BR BR BR BR
NS BI BI BI BI BI
e ZE MI MI MI MI MI
PS SM SM SM SM SM
PB SR SR SR SR SR

The calculation process of interval type-2 fuzzy controller is described as follows:

(1) Calculate the membership interval [u Aem (e), u Aem (e)] of the fuzzy input variable e and
.
.
.
the membership interval [u Ben e , u Ben e ] of the fuzzy input variable e.
.
(2) Calculate the activation degree interval Fmn (e,e) of each rule. The computational
formula is as follows:
.
.
.

h i
Fmn e, e = u Aem (e) × u Ben e , u Aem (e) × u Ben e = (ω Lmn , ωUmn ) (14)

where ω Lmn is the lower edge of the activate interval for the number mn rule; ωUmn is
the upper edge of the activate interval for the number mn rule. “×” stands for the
minimum value t-norm.
Step 3: type reduction
Sustainability 2021, 13, 11531 11 of 23

The activation degree interval of each rule is integrated with the membership function
of the output variable, and the KM algorithm [29] is used to simplify the type of fuzzy set
output obtained by integrated. The result of fuzzy inference can be expressed as

Ycos x0 = [ Tb_il , Tb_ir ]

(15)
RL Rg
f xuC ( x )dx + L xuC ( x )dx
Tb_il = RL Rg
uC ( x )dx + uC ( x )dx
R Rf RLg (16)
f xuC ( x )dx + R xuC ( x )dx
Tb_ir = RR Rg
f uC ( x )dx + R uC ( x )dx

where Tb_il and Tb_ir represent the left and right endpoints of the interval type-2 fuzzy
set output; f and g are the upper and lower bounds of the interval type-2 fuzzy output
interval in the fuzzy set domain, respectively; L is the left transition point, and R is the
right transition point.
Step 4: defuzzification
The final ideal anti-lock braking torque Tb_i can be calculated as follows:

Tb_il + Tb_ir
Tb_i = (17)
2

3.3. Allocation Strategies of Anti-Lock Braking Wheel Cylinder Pressure

In this study, the electro-hydraulic compound anti-lock braking control aims to realize
the maximum energy recovery efficiency on the premise of safe braking, so the ideal
Sustainability 2021, 13, x FOR PEER REVIEW 12 of 24
anti-lock braking torque Tb_i should be ensured that the regenerative braking torque
is always equal to the maximum energy recovery regenerative braking torque Tri_max
and the rest braking torque provided by hydraulic braking system under every kind of
the rest conditions
braking braking torque provided
except when by thehydraulic brakingbraking
ideal anti-lock system under
torqueevery
Tb_i iskind of
smaller than
braking conditions except when the ideal anti-lock braking torque T b_i is smaller than the
the motor’s maximum energy recovery regenerative braking torque Tri_max . Furthermore,
motor’s maximum energy recovery regenerative braking torque Tri_max. Furthermore,
when the ideal anti-lock braking torque Tb_i is smaller than the maximum energy recovery
when the ideal anti-lock braking torque Tb_i is smaller than the maximum energy recov-
regenerative braking torque Tri_max , the pattern of braking is switched from composite
ery regenerative braking torque Tri_max, the pattern of braking is switched from compo-
braking to pure
site braking regenerative.
to pure regenerative.
The proposed allocation
The proposed allocation strategies
strategies flowchart
flowchart is described
is described in Figure
in Figure 8. 8.

Begin

Y N
Tb_i≥Tri_max

Tr_i=Tri_max Tr_i=Tb_i
Th_i=Tb_i-Tri_max Th_i=0

VCU

Tr_i Th_i

Regenerative Tr_r Th_r Hydraulic

braking system braking system
Electric
vehicle wheels

Figure 8.
Figure 8. Flowchart
Flowchartofofthe braking
the torque
braking distribution
torque strategy.
distribution strategy.

VCU adjusts the liquid output solenoid valve current of the wheel cylinder accord-
ing to Equation (18) based on the difference between ideal hydraulic braking torque Th_i
and the real hydraulic braking torque Th_r.

Th_i  Th_r  0, increase current

Sustainability 2021, 13, 11531 12 of 23

VCU adjusts the liquid output solenoid valve current of the wheel cylinder according
to Equation (18) based on the difference between ideal hydraulic braking torque Th_i and
the real hydraulic braking torque Th _r .

 Th_i − Th_r < 0, increase current
T − Th_r = 0, maintain current (18)
 h_i
Th_i − Th_r > 0, decrease current

Thereby, the VCU controls the regenerative braking system working according to the
ideal regenerative braking torque Tr_i and produces real regenerative braking torque Tr_r .

4. Simulation Results and Discussion

In this section, simulations are conducted to demonstrate the control effects of the
proposed electro-hydraulic compound anti-lock braking control on the basis of interval
type-2 fuzzy logic anti-lock braking control method. The MATLAB/Simulink software
is widely used to simulate the real process of car driving [30]. Table 2 shows the main
parameters of the electric vehicle.

Table 2. Main parameters.

Variables Values Variables Values

m (kg) 960 δ 0
A (m2 ) 2.57 Pr (MPa) 0.375
Iz (m·kg2 ) 1600 Rω (m) 0.29
Pm (MPa) 15 Jω (m·kg2 ) 2.1

To verify the effectiveness of the proposed interval type-2 fuzzy logic control, the
braking performance is compared using two controllers as follows:
Electro-hydraulic compound anti-lock braking controller 1: the controller is applied
with an interval type-2 fuzzy logic control algorithm.
Electro-hydraulic compound anti-lock braking controller 2: the controller is applied
with a type-1 fuzzy logic control algorithm.
The initial vehicle velocity is set to 108 km/h. The ABS control quit to work and
increase the braking torque to a certain value to make the wheels locked when the vehicle
velocity decelerates to 10 km/h. Furthermore, the control precision of slip rate utilizes the
root-mean-square (RMS) of slip rate error to be quantified. The error is defined as follows:
Z t
eλi = (λi − λd )dt (19)
0

where eλi indicates the error of slip rate.

Two working conditions are simulated in this research to demonstrate the control
effects of the interval type-2 fuzzy logic anti-lock braking control method.
Working condition 1: the joint-µ road, which means the road peak friction coefficient
suddenly changed from the high value to low value at 2 s, which is referenced from [31].
Working condition 2: the split-µ road, which means the left side of the road is dry
with the high friction coefficient and the right side is snow with a low coefficient, which is
referenced from [32].
Table 3 is the parameters of dry road surface and snow road surface referenced
from [10].
Sustainability 2021, 13, 11531 13 of 23

Table 3. Road surface parameters.

Slip Rate Coefficient Coefficient

Road Condition
(Optimal) (Maximum) (Slip Rate = 1)
Dry 0.18 0.90 0.80
Snow 0.14 0.30 0.20

Sustainability 2021, 13, x FOR PEER REVIEW 14 of 24

4.1. Braking
Sustainability 2021, 13, x FOR PEER REVIEW
Performance Comparison under the Joint-µ Road Surface 14 of 24
Figure 9 shows the curve for the change of road peak adhesion coefficient µ. The µ
alters from 0.9 of dry road to 0.3 of snow road at 2 s.

Figure 9. The change of road peak adhesion coefficient under condition 1.

Thechange
Figure9.9.The
Figure changeofofroad
roadpeak
peakadhesion
adhesioncoefficient
coefficientunder
undercondition
condition1.1.

Figures 10–12exhibit
Figures 10–12 exhibitthe
thecomparison
comparison curves
curves of slip
of slip rate,rate,
totaltotal braking
braking torque,
torque, re-
regener-
Figures
generative 10–12
braking
ative braking exhibit
torque,
torque, the comparison
hydraulic
hydraulic curves
braking
braking of
torque,
torque, slip rate, total
and velocities
and velocities braking
underunder torque,
condition
condition 1. re-
1.
generative braking torque, hydraulic braking torque, and velocities under condition 1.

(a) (b)
(a) (b)
Figure 10.
Figure 10. The
The slip
slip rate
rate under
under condition 1: (a)
condition 1: (a) the
the slip
slip rate
rate of
of front wheels; (b)
front wheels; (b) the
the slip
slip rate
rate of
of rear
rear wheels.
wheels.
Figure 10. The slip rate under condition 1: (a) the slip rate of front wheels; (b) the slip rate of rear wheels.
Sustainability 2021, 13, 11531 (a) (b) 14 of 23

Figure 10. The slip rate under condition 1: (a) the slip rate of front wheels; (b) the slip rate of rear wheels.

Sustainability 2021, 13, x FOR PEER REVIEW 15 of 24

(a) (b)

(c) (d)

(e) (f)
Figure
Figure 11.11.
TheThe braking
braking torques
torques under
under condition
condition 1: (a)1:the
(a)total
the braking
total braking
torquetorque
of frontofwheels;
front wheels; (b) the
(b) the total totaltorque
braking braking
of rear wheels; (c) the regenerative braking torque of front wheels; (d) the regenerative braking torque of rear wheels; (e)ofthe
torque of rear wheels; (c) the regenerative braking torque of front wheels; (d) the regenerative braking torque rear
wheels; (e) the hydraulic braking torque of front wheels; (f) the hydraulic braking
hydraulic braking torque of front wheels; (f) the hydraulic braking torque of rear wheels. torque of rear wheels.
 (e) (f)
Figure2021,
Sustainability 11. 13,
The braking torques under condition 1: (a) the total braking torque of front wheels; (b) the total braking
11531 15 of 23
torque of rear wheels; (c) the regenerative braking torque of front wheels; (d) the regenerative braking torque of rear
wheels; (e) the hydraulic braking torque of front wheels; (f) the hydraulic braking torque of rear wheels.

(a) (b)
Figure12.
Figure 12.The
Thevehicle
vehicleand
andwheels
wheelsvelocity
velocityfor
fortwo
twocontrollers
controllersunder
undercondition
condition1:1:(a)
(a)the
thevehicle
vehicleand
andwheels
wheelsvelocity
velocityfor
for
controller 1; (b) the vehicle and wheels velocity for controller
controller 1; (b) the vehicle and wheels velocity for controller 2.2.

As shown
As shown in Figure
Figure 10,
10,the
theRMS
RMSofofslip rate
slip error
rate of front
error andand
of front rearrear
wheels withwith
wheels con-
troller 1 are
controller reduced
1 are by 54.44%
reduced and 57.28%
by 54.44% compared
and 57.28% to controller
compared 2, respectively,
to controller which
2, respectively,
which illustrates the control effects of the proposed interval type-2 fuzzy logic strategy
with the change of peak adhesion coefficient have smaller fluctuations than type-1 fuzzy
logic. Moreover, there is an abrupt jitter at 2 s where the road surface changed for both two
controllers, but the degree of jitter of controller 1 is lower than controller 2, which indicates
the slip rate control transition of interval type-2 fuzzy control is more stable than type-1
fuzzy logic at the time of 2 s when the road surface suddenly altered and illustrates the
interval type-2 fuzzy logic anti-lock braking control has better anti-interference ability than
type-1 fuzzy logic anti-lock braking control. Therefore, the interval type-2 fuzzy logic ABS
control not only has strong robustness against uncertainties in road adhesion coefficient
but also achieves an outstanding slip rate control for the vehicle.
Figure 11 exhibits the total braking torque and braking torque of subsystems. The total
braking torque and hydraulic braking torque of controller 1 have smaller fluctuations than
that of controller 2 when tracking the ideal slip rate. Moreover, the control of torque could
maintain stability when the road surface changed and the vertical load was transferred
in the braking process, which indicates the interval type-2 fuzzy logic control has better
performances when it confronts external and internal interference.
Figure 12 shows the velocity change of vehicle and wheels. The vehicle velocity is
close to the wheel velocity, and the dangerous situation of locked does not occur during
the braking process, which means better effect of slip control, and the wheel’s velocity
variation of controller 1 have low-frequency jitter than that of controller 2 during the whole
ABS control, which demonstrates the smaller fluctuations of regenerative braking torque
of controller 1. Figure 13 exhibits the curves of the vehicle’s kinetic energy and reclaimed
regenerative braking energy, and the energy recovery efficiency could reach 7.6%, which
illustrates better energy recovery efficiency of the electric vehicle under a joint-µ road.
 the braking process, which means better effect of slip control, and the wheel’s velocity
variation of controller 1 have low-frequency jitter than that of controller 2 during the
whole ABS control, which demonstrates the smaller fluctuations of regenerative braking
torque of controller 1. Figure 13 exhibits the curves of the vehicle’s kinetic energy and
Sustainability 2021, 13, 11531
reclaimed regenerative braking energy, and the energy recovery efficiency could16reachof 23
7.6%, which illustrates better energy recovery efficiency of the electric vehicle under a
joint-μ road.

Figure 13. The

Figure 13. The energy
energy change
change of
of vehicle
vehicle with
with controller
controller 11 under
under condition
condition 1.
1.
4.2. Braking Performance Comparison under the Spilt-µ Road Surface
4.2. Braking Performance Comparison under the Spilt-μ Road Surface
Figure 14 exhibits the peak adhesion coefficient µ of joint-µ road, the right side of the
Figure 14 exhibits the peak adhesion coefficient μ of joint-μ road, the right side of
Sustainability 2021, 13, x FOR PEER REVIEWwith high value of friction coefficient while the left side with a low value of friction
road 17 of 24
the road with
coefficient. Thishigh valuecondition
working of friction coefficientthe
aggravates while the left
uncertain side with aoflow
environment the value
brakingof
friction coefficient. This working condition aggravates the uncertain environment of the
process.
braking process.

Figure
Figure 14.
14. The
The road
road peak
peak adhesion
adhesion coefficient
coefficient of
of condition
condition 2.

Figures 15–19
Figures 15–19 exhibit
exhibitthe
thecomparison
comparisoncurves of of
curves slipslip
rate, totaltotal
rate, braking torque,
braking regener-
torque, re-
ative braking torque, hydraulic braking torque, and velocity between controllers 1 and 2
generative braking torque, hydraulic braking torque, and velocity between controllers 1
under condition 2.
and 2 under condition 2.

(a) (b)
 Figure 14. The road peak adhesion coefficient of condition 2.

Sustainability 2021, 13, 11531 Figures 15–19 exhibit the comparison curves of slip rate, total braking torque, re- 17 of 23
generative braking torque, hydraulic braking torque, and velocity between controllers 1
and 2 under condition 2.

(a) (b)

Sustainability 2021, 13, x FOR PEER REVIEW 18 of 24

As shown in Figure 15, all the controllers could remain optimal slip rate tracking;
however, the RMS of slip rate error for each wheel of controller 1 is reduced by 33.92%,
(c) 67.61%, 28.27%, and 46.30%, respectively. The slip control curves (d) of interval type-2 fuzzy
logic have 2:
smaller fluctuations thanright
thatwheel;
of type-1 fuzzy logicofbefore 4 s,wheel;
which(c)illus-
Figure 15.Figure 15. The
The slip rateslip rate under
under condition
condition 2: (a) (a)
thethe sliprate
slip rateof
of front
front right wheel;(b)(b)
the slip slip
the rate front
rate of left
front the (c) the
left wheel;
slip rate of rear right wheel; (d)trates the control
the slip effect
rate of rear leftofwheel.
interval type-2 fuzzy logic with the different road surfaces for
slip rate of rear right wheel; (d) thewheels slip rate of rear
better than left wheel.
type-1 fuzzy logic and preferable adaption of diverse working condi-
tions.

(a) (b)

(c) (d)
Figure 16. The total braking torques under condition 2: (a) the total braking torque of front right wheel; (b) the total
Figure 16. The total braking torques under condition 2: (a) the total braking torque of front right wheel; (b) the total braking
braking torque of front left wheel; (c) the total braking torque of rear right wheel; (d) the total braking torque of rear left
torque of front left wheel; (c) the total braking torque of rear right wheel; (d) the total braking torque of rear left wheel.
wheel.

Figures 16–18 illustrate the braking torque variation of controller 1 are more stable
than that of controller 2 when the right wheels are braking on high friction coefficient
and the left are braking on low friction coefficient. Due to the too small wheels velocity,
the fluctuations of hydraulic braking torque become larger; however, the vehicle veloci-
ty has already reached to a low value, which means the fluctuations have less impact on
the braking safety.
Sustainability 2021, 13,
Sustainability 2021, 13, 11531
x FOR PEER REVIEW 18
19 of 23
24
Sustainability 2021, 13, x FOR PEER REVIEW 19 of 24

(a)(a) (b)
(b)

(c)(c) (d)
(d)
Figure 17.17. The
The regenerative
regenerative braking
braking torques
torques under
under condition
condition 2: 2: (a) the
(a)(a)
the regenerative
regenerative braking
braking torque of front right wheel;
Figure
Figure 17. The regenerative braking torques under condition 2: the regenerative brakingtorque
torqueofoffront right
front right wheel;
wheel;
(b) the
(b)(b) regenerative
thethe
regenerative braking torque of front left wheel; (c) the regenerative braking torque of rear right wheel; (d) thethe
re-
regenerativebraking
brakingtorque
torqueofoffront
frontleft
leftwheel;
wheel;(c)
(c) the
the regenerative
regenerative braking torque
torque of
of rear
rear right
rightwheel;
wheel;(d)(d)the re-
generative braking torque of rear left wheel.
generative braking
regenerative braking torque
torque of
of rear
rear left
left wheel.
wheel.

(a)(a) (b)
(b)
Figure 18. Cont.
Sustainability 2021,13,
Sustainability2021, 13,11531
x FOR PEER REVIEW 1920ofof2324

(c) (d)
Figure 18. The hydraulic braking torques under condition 2: (a) the hydraulic braking torque of front right wheel; (b) the
hydraulic braking torque of front left wheel; (c) the hydraulic braking torque of rear right wheel; (d) the hydraulic brak-
ing torque of rear left wheel.

Figure 19 exhibits the velocity of the vehicle and wheels. The velocity variation of
the rear left wheel for the two controllers are similar under a low value of friction coeffi-
cient refer to wet road. However, the vehicle front right wheel velocity of controller 1
has less jitters than that of controller 2 under a high value of friction coefficient, which
(c) means better braking safety and braking comfort ability (d)when the wheels braking on
Figure18.
Figure 18.The
Thehydraulic different
hydraulicbraking
brakingtorquessurfaces
torques under simultaneously.
undercondition
condition2:2:(a)
(a)theThe
the results
hydraulic
hydraulic illustrate
braking
braking the
torque
torque interval
ofoffront
frontrighttype-2
right wheel;fuzzy
wheel; thelogic
(b)the
(b)
hydraulicbraking
hydraulic brakingtorque anti-lock
torqueofof front
front left
left braking
wheel;
wheel; control
(c)(c)
thethe has braking
hydraulic
hydraulic better anti-interference
braking torque
torque of rear
of rear ability
right
right and(d)
wheel;
wheel; (d) thebetter
the adaption
hydraulic
hydraulic of dif-
brak-
braking
ing torque of rear left
torque of rear left wheel. ferent
wheel. working conditions than the traditional type-1 fuzzy logic control.

Figure 19 exhibits the velocity of the vehicle and wheels. The velocity variation of
the rear left wheel for the two controllers are similar under a low value of friction coeffi-
cient refer to wet road. However, the vehicle front right wheel velocity of controller 1
has less jitters than that of controller 2 under a high value of friction coefficient, which
means better braking safety and braking comfort ability when the wheels braking on
different surfaces simultaneously. The results illustrate the interval type-2 fuzzy logic
anti-lock braking control has better anti-interference ability and better adaption of dif-
ferent working conditions than the traditional type-1 fuzzy logic control.

(a) (b)
Figure 19.
Figure 19. The
The vehicle
vehicle and
and wheel
wheel velocities
velocities for
for two
two controllers
controllers under
under condition
condition 2:
2: (a)
(a) the
the vehicle
vehicle and
and wheel
wheel velocities
velocities for
for
controller 1; (b) the vehicle and wheel velocities for controller
controller 1; (b) the vehicle and wheel velocities for controller 2.2.

Figure
As shown20 exhibits
in Figure the15,
curves ofcontrollers
all the vehicle’s kinetic
couldenergy
remainand reclaimed
optimal regenerative
slip rate tracking;
braking energy.
however, the RMS In of
Figure 20, error
slip rate the energy recovery
for each wheel of efficiency could
controller 1 is reach
reduced 9.38%, which
by 33.92%,
illustrates
67.61%, betterand
28.27%, energy recovery
46.30%, efficiency
respectively. of an
The slipelectric
controlvehicle
curvesunder a split-μ
of interval road.
type-2 fuzzy
logic have smaller fluctuations than that of type-1 fuzzy logic before 4 s, which illustrates
the control effect of interval type-2 fuzzy logic with the different road surfaces for wheels
(a)
better than type-1 fuzzy logic and preferable adaption of diverse (b) working conditions.
Figure 19. The vehicle and wheel Figures 16–18
velocities illustrate
for two the braking
controllers torque 2:
under condition variation of controller
(a) the vehicle and wheel 1 are more for
velocities stable
controller 1; (b) the vehiclethan that of
and wheel controller
velocities 2 when the
for controller 2. right wheels are braking on high friction coefficient
and the left are braking on low friction coefficient. Due to the too small wheels velocity,
the fluctuations of hydraulic
Figure 20 exhibits braking
the curves oftorque become
vehicle’s kineticlarger;
energyhowever, the vehicle
and reclaimed velocity
regenerative
has already
braking reached
energy. In to a low20,
Figure value,
the which
energymeans the fluctuations
recovery havereach
efficiency could less impact
9.38%,on the
which
braking safety.
illustrates better energy recovery efficiency of an electric vehicle under a split-μ road.
Figure 19 exhibits the velocity of the vehicle and wheels. The velocity variation of the
rear left wheel for the two controllers are similar under a low value of friction coefficient
refer to wet road. However, the vehicle front right wheel velocity of controller 1 has less
jitters than that of controller 2 under a high value of friction coefficient, which means
better braking safety and braking comfort ability when the wheels braking on different
Sustainability 2021, 13, 11531 20 of 23

surfaces simultaneously. The results illustrate the interval type-2 fuzzy logic anti-lock
braking control has better anti-interference ability and better adaption of different working
conditions than the traditional type-1 fuzzy logic control.
Sustainability 2021, 13, x FOR PEER REVIEW Figure 20 exhibits the curves of vehicle’s kinetic energy and reclaimed regenerative
21 of 24
braking energy. In Figure 20, the energy recovery efficiency could reach 9.38%, which
illustrates better energy recovery efficiency of an electric vehicle under a split-µ road.

Figure 20.
Figure 20. The energy change
change of
of vehicle
vehicle with
with controller
controller 11 under
under condition
condition 2.
2.

5.
5. Conclusions
Conclusions
In
In this
this present
present study,
study, an an efficient
efficient interval
interval type-2
type-2 fuzzy
fuzzy logic
logic control
control method
method is is de-
de-
veloped and successfully applied in an electro-hydraulic compound
veloped and successfully applied in an electro-hydraulic compound anti-lock braking anti-lock braking
controller.
controller. The The interval
interval type-2
type-2 fuzzy
fuzzy logic
logic enhances
enhances the the ability
ability ofof anti-interference
anti-interference for for
fuzzy control during the braking process by the membership
fuzzy control during the braking process by the membership function function expansion
expansion method.
meth-
It
od.means
It meansevery single
every fuzzy
single variable
fuzzy is described
variable by both
is described lower
by both and and
lower upper membership
upper member-
functions. Furthermore,
ship functions. to fullytoexert
Furthermore, fullythe advantages
exert of regenerative
the advantages braking and
of regenerative recapture
braking and
more lost kinetic
recapture more lostenergy, theenergy,
kinetic corresponding braking torque
the corresponding allocation
braking torquestrategy is designed
allocation strategy
to maintain the maximum energy recovery efficiency on the premise of safe braking. Ad-
is designed to maintain the maximum energy recovery efficiency on the premise of safe
ditionally, to verify the braking performances of the proposed controller, simulations of
braking. Additionally, to verify the braking performances of the proposed controller,
two working conditions specified to complex and changing road surface are conducted
simulations of two working conditions specified to complex and changing road surface
to illustrate the effectiveness of the proposed controller compared with type-1 fuzzy logic
are conducted to illustrate the effectiveness of the proposed controller compared with
control. The simulation results show that the wheel slip rate tracking RMS of interval
type-1 fuzzy logic control. The simulation results show that the wheel slip rate tracking
type-2 fuzzy logic controller for front and rear wheels decreased by 54.44% and 57.28%,
RMS of interval type-2 fuzzy logic controller for front and rear wheels decreased by
respectively, compared with type-1 fuzzy logic control under the joint-µ road, and for each
54.44% and 57.28%, respectively, compared with type-1 fuzzy logic control under the
wheel, the RMS reduced by 33.92%, 67.61%, 28.27%, and 46.30%, compared with type-1
joint-μ road, and for each wheel, the RMS reduced by 33.92%, 67.61%, 28.27%, and
fuzzy logic control under the spilt-µ road. Moreover, ideal regenerative braking energy
46.30%, compared with type-1 fuzzy logic control under the spilt-μ road. Moreover, ide-
recovery efficiencies of 7.6% and 9.38% can be achieved by the proposed electro-hydraulic
al regenerative braking energy recovery efficiencies of 7.6% and 9.38% can be achieved
compound anti-lock braking system under joint-µ and split-µ road surfaces, respectively.
byaddition,
In the proposed electro-hydraulic
the braking compound
torque variation curveanti-lock
of type-2braking systemisunder
fuzzy control more joint-μ and
stable and
split-μ road surfaces, respectively. In addition, the braking torque
with minor fluctuation than type-1 fuzzy control, which indicates the better robustness of variation curve of
type-2 fuzzy
braking torquecontrol
controlisin
more stable and
the process with minor
of anti-lock fluctuation
braking and bring than type-1 fuzzy
preferable brakingcontrol,
feel-
which indicates the better robustness of braking torque control
ing to the driver. Therefore, the interval type-2 fuzzy logic control ensures the functionin the process of an-of
ti-lock braking
energy recoveryand andbring preferable
anti-lock brakingbraking feeling
with fully to the driver.
improving Therefore, thestability,
the anti-interference, interval
type-2
and fuzzy logic
robustness control
of slip ensuresinthe
rate control thefunction
uncertainof process
energy of recovery and anti-lock
electro-hydraulic braking
compound
with fully improving
anti-lock braking. the anti-interference, stability, and robustness of slip rate control in
the uncertain
In furtherprocess
research,ofour
electro-hydraulic
future concentrated compound
work isanti-lock braking. the interval type-
about improving
In further research, our future concentrated
2 fuzzy controller from two main aspects. Innovative braking torquework is about improving
allocationthe interval
strategy is
type-2 fuzzy controller from two main aspects. Innovative braking
a key issue to solve, which is useful for interval type-2 fuzzy logic controllers to obtain a torque allocation
strategy is a key issue to solve, which is useful for interval type-2 fuzzy logic controllers
to obtain a better braking energy recovery. Furthermore, a less time-consuming type
reduction method would be studied to optimize the interval type-2 fuzzy logic algo-
rithm.
Sustainability 2021, 13, 11531 21 of 23

better braking energy recovery. Furthermore, a less time-consuming type reduction method
would be studied to optimize the interval type-2 fuzzy logic algorithm.

Author Contributions: Conceptualization, L.L. and J.W.; methodology, L.L.; software, L.L.; valida-
tion, L.L. and J.W.; formal analysis, L.L.; investigation, J.W.; resources, J.L.; data curation, L.L. and J.L.;
writing—original draft preparation, L.L.; writing—review and editing, J.W. and J.L.; visualization,
J.L.; supervision, J.W. All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by the Zhejiang Provincial Department of Education Scientific
Research Project (Grant No. 20020060-F); Open Research Fund of Anhui Engineering Technology
Research Center of Automotive New Technique (Grant No. QCKJ202102); Wenzhou Major Science
and Technology Innovation Project of China (ZG2021028); and Scientific Research Start-up Project at
Zhejiang Sci-Tech University (Grant No. 20022303-Y).
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement: Data are available upon request by contacting the first author.
Acknowledgments: Thank you to all reviewers for their suggestions to make this paper better.
Conflicts of Interest: The authors declare no conflict of interest.

Nomenclature

A windward area of vehicle Rω rolling radius of wheel

A
e type-2 fuzzy set of e RΩ internal resistance of motor
Af area of brake wheel cylinder piston SH horizontal bias of Magic Formula tire model
B stiffness coefficient of Magic Formula tire model SV vertical bias of Magic Formula tire model
.
B
e type-2 fuzzy set of e Tb braking torque
C shape coefficient of Magic Formula tire model Tb_i ideal anti-lock braking torque
C
e type-2 fuzzy set of Tb_i Tb_il left endpoint of the interval type-2 fuzzy set output
equivalent liquid capacity characteristic coefficient
Ce Tb_ir right endpoint of the interval type-2 fuzzy set output
of the pipeline and wheel cylinder
c the center position of the membership function Tb_r real anti-lock braking torque
D peak value of Magic Formula tire model Td driving torque
d d-axis of permanent magnet synchronous motor Tf rolling resistance torque
E curvature coefficient of Magic Formula tire model Th hydraulic braking torque
difference value of wheel slip rate and the ideal
e Th_i ideal hydraulic braking torque
slip rate
.
e change rate of e Th_r real hydraulic braking torque
eλi slip rate error Tr regenerative braking torque
Fx longitudinal force Tr_i ideal regenerative braking torque
Fy lateral force Tr_r real regenerative braking torque
FD air resistance force Tri_maxmaximum energy recovery regenerative braking torque
Ff rolling resistance force tf front wheelbase
Fmn activation degree interval of mn rule tr rear wheelbase
upper bound of the interval type-2 fuzzy output
f u second variables of type-2 fuzzy set
interval
lower bound of the interval type-2 fuzzy output
g ud voltage motor d-axis
interval
Iz vehicle’s moment of inertia around the z-axis uq voltage motor q-axis
ia actual current of phase a vx velocity of electric vehicle along the x-axis
ib actual current of phase b vy velocity of electric vehicle along the y-axis
.
ic actual current of phase c vx acceleration of electric vehicle along the x-axis
.
iai expected current of phase a vy acceleration of electric vehicle along the y-axis
ibi expected current of phase b X fuzzy domain
ici expected current of phase c x first variable of type-2 fuzzy set
id current of motor d-axis Y output variable of Magic Formula tire model
Sustainability 2021, 13, 11531 22 of 23

iq current of motor q-axis β angle between the air resistance and the driving direction
idi expected current of motor d-axis γ vehicle yaw angle
.
iqi expected current of motor q-axis γ yaw angular velocity
Jx primary membership degree δ steering angle of the wheel
Jw wheel’s moment of inertia η wheel cylinder efficiency
L left transition point θ rotation angle of rotor
Ld inductor of motor d-axis λ slip rate
Lq inductor of motor q-axis λd ideal slip rate
distances between the mass center of the vehicle
la µ road peak adhesion coefficient
and the front axle
distances between the mass center of the vehicle
lb µb friction coefficient of brake
and the rear axle
m the mass of the electric vehicle µ Ae secondary membership degree
Pm pressure of the main cylinder ψf motor magnetic chain
Pr pressure of low-pressure accumulator τ time delay of brake
transmission lag time of solenoid valve and pipeline during
Pw pressure of the wheel cylinder τvp
pressurization
0 transmission lag time of solenoid valve and pipeline during
p number of pole-pairs τvp
decompression
q q-axis of permanent magnet synchronous motor ω angular velocity of wheel
.
R the right transition point ω angular acceleration of the wheel
Rb brake effective radius of friction ωd rotor angular velocity of motor
equivalent liquid resistance characteristic
Re coefficient of the pipeline and wheel cylinder ω Lmn lower edge of the activate interval for number mn rule
when the pressure is increased
equivalent liquid resistance characteristic
R0e coefficient of the pipeline and wheel cylinder ωUmn upper edge of the activate interval for number mn rule
when the pressure is reduced

References
1. Hu, D.H.; Wang, Y.T.; Li, J.W.; Yang, Q.Q.; Wang, J. Investigation of optimal operating temperature for the PEMFC and its tracking
control for energy saving in vehicle applications. Energy Convers. Manag. 2021, 249, 114842. [CrossRef]
2. Liu, B.H.; Li, L.; Wang, X.Y.; Cheng, S. Hybrid electric vehicle downshifting strategy based on stochastic dynamic programming
during regenerative braking process. IEEE Trans. Veh. Technol. 2018, 67, 4716–4727. [CrossRef]
3. Ko, S.; Song, C.; Kim, H. Cooperative control of the motor and the electric booster braking to improve the stability of an in-wheel
electric vehicle. Int. J. Automot. Technol. 2016, 17, 447–456. [CrossRef]
4. Pretagostini, F.; Ferranti, L.; Berardo, G.; Ivanov, V.; Shyrokau, B. Survey on wheel slip control design strategies, evaluation and
application to antilock braking systems. IEEE Access 2020, 8, 10951–10970. [CrossRef]
5. Yuan, Y.; Zhang, J.Z.; Li, Y.T.; Li, C. A novel regenerative electrohydraulic brake system: Development and hardware-in-loop
tests. IEEE Trans. Veh. Technol. 2018, 67, 11440–11452. [CrossRef]
6. Rajendran, S.; Spurgeon, S.K.; Tsampardoukas, G.; Hampson, R. Estimation of road frictional force and wheel slip for effective
anti-lock braking system (ABS) control. Int. J. Robust Nonlinear Control 2019, 29, 736–765. [CrossRef]
7. Chiang, W.P.; Yin, D.J.; Shimizu, H. Slip-based regenerative ABS control for in-wheel-motor drive EV. J. Chin. Inst. Eng. 2015, 38,
220–231. [CrossRef]
8. Yang, Y.; Tang, Q.S.; Bolin, L.; Fu, C.Y. Dynamic coordinated control for regenerative braking system and anti-lock braking system
for electrified vehicles under emergency braking conditions. IEEE Access 2020, 8, 172664–172667. [CrossRef]
9. Feng, X.L.; Hu, J. Discrete fuzzy adaptive PID control algorithm for automotive anti-lock braking system. J. Ambient. Intell.
Humaniz. Comput. 2021, 1–10. [CrossRef]
10. Sun, J.H.; Xue, X.D.; Cheng, K.W.E. Fuzzy sliding mode wheel slip ratio control for smart vehicle anti-lock braking system.
Energies 2019, 12, 2501. [CrossRef]
11. Verma, R.; Ginoya, D.; Shendge, P.D.; Phadke, S.B. Slip regulation for anti-lock braking system using multiple surface sliding
controller combined with inertial delay control. Veh. Syst. Dyn. 2015, 53, 1150–1171. [CrossRef]
12. Nguyen, A.-T.; Taniguchi, T.; Eciolaza, L.; Campos, V.; Palhares, R.; Sugeno, M. Fuzzy control systems: Past, present and future.
IEEE Comput. Intell. Mag. 2019, 14, 56–58. [CrossRef]
13. Wang, T.C.; Sui, S.; Tong, S.C. Data-based adaptive neural network optimal output feedback control for nonlinear systems with
actuator saturation. Neurocomputing 2017, 247, 192–201. [CrossRef]
14. Banaei, A.; Alamatian, J. New genetic algorithm for structural active control by considering the effect of time delay. J. Vib. Control
2021, 27, 743–758. [CrossRef]
Sustainability 2021, 13, 11531 23 of 23

15. Fargione, G.; Tringali, D.; Risitano, G. A fuzzy-genetic control system in the ABS for the control of semi-active vehicle suspensions.
Mechatronics 2016, 3, 89–102. [CrossRef]
16. Aksjonov, A.; Vodovozov, V.; Augsburg, K.; Petlenkov, E. Design of regenerative anti-lock braking system controller for 4
in-wheel-motor drive electric vehicle with road surface estimation. Int. J. Automot. Technol. 2018, 19, 727–742. [CrossRef]
17. Mokarram, M.; Khoei, A.; Hadidi, K. A fuzzy Anti-lock braking system (ABS) controller using CMOS circuits. Microprocess.
Microsyst. 2019, 70, 47–52. [CrossRef]
18. Castillo, O.; Amador-Angulo, L.; Castro, J.R.; Garcia-Valdez, M. A comparative study of type-1 fuzzy logic systems, interval
type-2 fuzzy logic systems and generalized type-2 fuzzy logic systems in control problems. Inf. Sci. 2016, 354, 257–274. [CrossRef]
19. Hailemichael, A.; Salaken, S.M.; Karimoddini, A.; Homaifar, A.; Abbas, K.; Nahavandi, S. Developing a computationally effective
interval type-2 TSK fuzzy logic controller. J. Intell. Fuzzy Syst. 2020, 38, 1915–1928. [CrossRef]
20. Gonzalez, C.I.; Melin, P.; Castro, J.R.; Mendoza, O.; Castillo, O. An improved sobel edge detection method based on generalized
type-2 fuzzy logic. Soft Comput. 2016, 20, 773–784. [CrossRef]
21. Zhang, Z.M. Trapezoidal interval type-2 fuzzy aggregation operators and their application to multiple attribute group decision
making. Neural Comput. Appl. 2018, 29, 1039–1054. [CrossRef]
22. Gaxiola, F.; Melin, P.; Valdez, F.; Castro, J.R.; Castillo, O. Optimization of type-2 fuzzy weights in backpropagation learning for
neural networks using Gas and PSO. Appl. Soft Comput. 2016, 38, 860–871. [CrossRef]
23. Sanchez, M.A.; Castillo, O.; Castro, J.R. Generalized type-2 fuzzy systems for controlling a mobile robot and a performance
comparison with interval type-2 and type-1 fuzzy systems. Expert Syst. Appl. 2015, 42, 5904–5914. [CrossRef]
24. Bin Peeie, M.H.; Ogino, H.; Oshinoya, Y. Skid control of a small electric vehicle with two in-wheel motors: Simulation model of
ABS and regenerative brake control. Int. J. Crashworthiness 2016, 21, 396–406. [CrossRef]
25. Yu, D.L.; Wang, W.S.; Zhang, H.B. Research on Anti-Lock Braking Control Strategy of Distributed-Driven Electric Vehicle. IEEE
Access 2020, 8, 162467–162478. [CrossRef]
26. Vignati, M.; Sabbioni, E. Force-based braking control algorithm for vehicles with electric motors. Veh. Syst. Dyn. 2020, 58,
1348–1366. [CrossRef]
27. Wang, J.C.; He, R.; Kim, Y.-B. Optimal Anti-Lock Braking Control with Nonlinear Variable Voltage Charging Scheme for an
Electric Vehicle. IEEE Trans. Veh. Technol. 2020, 69, 7211–7222. [CrossRef]
28. Zhao, X.M.; Mo, H.; Yan, K.F.; Li, L.X. Type-2 fuzzy control for driving state and behavioral decisions of unmanned vehicle.
IEEE-CAA J. Autom. Sin. 2020, 7, 178–186. [CrossRef]
29. Karnik, N.N.; Mendel, J.M.; Liang, Q.L. Type-2 fuzzy logic system. IEEE Trans. Fuzzy Syst. 1999, 7, 643–658. [CrossRef]
30. Wang, J.C.; Lv, L.F.; Ren, J.Y.; Chen, S.A. Time delay compensation using a Taylor series compound robust scheme for a semi-active
suspension with magneto rheological damper. Asian J. Control. 2021. [CrossRef]
31. Wang, J.C.; He, R.; Kim, Y.-B. Optimal control of regenerative hydraulic composite braking system based on a voltage variable
charging control scheme. Proc. Inst. Mech. Eng. Part D-J. Automob. Eng. 2020, 234, 536–551. [CrossRef]
32. Aksjonov, A.; Augsburg, K.; Vodovozov, V. Design and simulation of the robust ABS and ESP fuzzy logic controller on the
complex braking maneuvers. Appl. Sci. 2016, 6, 382. [CrossRef]