11440 IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 67, NO.

12, DECEMBER 2018

A Novel Regenerative Electrohydraulic Brake

System: Development and Hardware-in-Loop Tests
Ye Yuan , Junzhi Zhang, Yutong Li, and Chao Li

Abstract—The development of a novel electrohydraulic brake related research institutes. Energy consumption, function con-
system named REHB is presented in this paper. Neither high- figurations as well as pricing turn into important assessment
pressure accumulator nor linear valves are adopted to reduce criteria for production vehicles. New functions like RBS (Re-
the system cost, the difficulty of manufacturing, assembling,
and the risk of leakage. Stroke simulators are implemented to generative Braking System) and ADAS (Advanced Driver As-
elevate the quality of pedal feel. The braking force blending sistance Systems) have made higher demands for vehicle chas-
control strategy is well designed to coordinate regenerative and sis, especially for brake systems [1], [2]. These requirements
friction brake force, which has ensured maximized regenerative for improving brake systems can be generally categorized into
brake force. In the aspect of underlying control, current amplitude three aspects.
modulation control is adopted to improve the accuracy of
hydraulic pressure modulation and eliminate vibration noise.
High fidelity models of vehicle and brake system are built based A. Brake-by-Wire Characteristic
on MATLAB-AMESim co-simulation platform. Three demand
levels of one-step-brake scenarios are simulated. Simulation Compared with any other automotive optimization technique,
results show the reasonability of proposed hydraulic structure and RBS is the most efficient method in improving vehicle energy
designed control strategy. A regenerative brake test bench is set consumption according to recent researches [3]–[5]. However,
up based on a prototype of the developed system. Three identical the capacity of regenerative brake force varies depending on ve-
one-step-brake experiments are conducted on the test bench.
According to the bench test results, the blending of regenerative hicle velocity and battery efficiency. Hence, conventional fric-
and friction brake force tracks driver brake demand adequately tion brake system is indispensable on electrified vehicles, to
well. Regenerative efficiencies are calculated as 46.32%, 43.93%, compensate the varying of regenerative brake force [6]. Total
and 31.88%, respectively, in three bench tests. Compared with the actual brake force is the addition of regenerative and friction
previous generation, no obvious valve buzzing noise is observed, brake forces. Therefore, hydraulic brake should be regulated
and jerks are whittled significantly. REHB has provided a brake-
by-wire solution of safety, high energy recuperation efficiency independently with high accuracy based on driver’s request and
with low cost, and uncompromising performance. varying of regenerative brake force [7].
On the other hand, functions in ADAS, like ACC (Adaptive
Index Terms—Brake-by-wire, braking force blending control,
current amplitude modulation, electrohydraulic brake system,
Cruise Control) and AEB (Autonomous Emergency Braking)
regenerative brake. can perform braking actively without any driver braking demand
[8]. This feature can ensure sufficient distance with foregoing
I. INTRODUCTION vehicle or avoid collision with any object ahead.
ITH the development of novel technologies like AI (arti- To sum up, the hydraulic pressures in wheel cylinders are not
W ficial intelligence) and improvement of cognitions such
as environmental conservation, electrified and autonomous ve-
always same with pressure in master cylinder, which can only
be carried out by brake-by-wire characteristic. There are two
hicles have become the main focus of automotive industry and common solutions to achieve brake-by-wire mechanism. One
is high pressure accumulator based solution, and the other is
electric booster based [9].
Manuscript received February 23, 2018; revised June 16, 2018; accepted
September 7, 2018. Date of publication September 24, 2018; date of current
version December 14, 2018. This work was supported in part by the National B. Pedal-Feel Compensation
Key Research and Development Program of China under 2016YFB0101402.
The review of this paper was coordinated by Dr. S. Anwar. (Corresponding Pressure in master cylinder is not same with pressures in
author: Ye Yuan.) wheel cylinders, hence, it is essential to equip pedal feel simu-
Y. Yuan and C. Li are with the Department of Automotive Engineering, lators to compensate brake pedal force in brake-by-wire systems
Tsinghua University, Beijing 100084, China (e-mail:, yuan.ye.yerik@gmail.
com; [email protected]). [10]. Thus, drivers can feel similar brake pedal feel with conven-
J. Zhang is with Department of Automotive Engineering, Tsinghua Univer- tional hydraulic brake system, and won’t be aware of switching
sity, Beijing 100084, China and also with Collaborative Innovation Center of between functions. Namely, the driver’s pedal operation should
Electric Vehicles in Beijing, Beijing 100081, China (e-mail:, jzhzhang@mail.
tsinghua.edu.cn). be isolated (or decoupled) from hydraulic brake system [11].
Y. Li is with the Department of Transportation, Tongji University, Shanghai With corresponding to two brake-by-wire solutions mentioned
201801, China (e-mail:,[email protected]). above, pedal stroke simulator based and motor control compen-
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org. sation based are two mainly adopted pedal feel compensation
Digital Object Identifier 10.1109/TVT.2018.2872030 techniques [12].
0018-9545 © 2018 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications standards/publications/rights/index.html for more information.

Authorized licensed use limited to: Air University. Downloaded on May 31,2021 at 07:48:16 UTC from IEEE Xplore. Restrictions apply.
YUAN et al.: NOVEL REGENERATIVE ELECTROHYDRAULIC BRAKE SYSTEM: DEVELOPMENT AND HARDWARE-IN-LOOP TESTS 11441

TABLE I Nissan and Hitachi [15] cooperatively proposed an

CHINA ELECTRIC CAR SALES TOP 10 (2017)
electrically-driven brake-by-wire system e-ACT. In the e-ACT
system, ball-bolt-screw mechanism are designed to convert in-
ner motor rotational motion into linear motion, which is used
as the booster source. Hence, conventional vacuum booster is
eliminated. The braking characteristic can be adjusted and pedal
feel is kept natural and smooth by using a simple pedal-force
compensator. The e-ACT system was employed successfully in
electric vehicle Leaf manufactured by Nissan [16].
Continental developed an electronic hydraulic brake system
MK C1 [17]. The vacuum pump and pump motor in conven-
tional ABS system are removed, and motor booster unit, execu-
tion unit as well as pressure modulation unit are all converged
into one compact assembly. Compared with conventional hy-
draulic system, MK C1 can actively build up braking pressure
faster, which can satisfy pressure modulation requirements of
various brake functions including ESC, ABS and regenerative
braking.
TABLE II
COMPARISON BETWEEN SYSTEMS
Bosch [18] adopted a similar technique with Continental and
invented a novel type of vacuum-free by-wire brake system,
which has been successfully implemented on Tesla electric ve-
hicles [19], i.e., iBooster. The vacuum booster is replaced with
motor and rectilinear motion conversion mechanism, while pres-
sure modulation module remains the same with conventional
ESP. iBooster can fulfill active deceleration up to 0.3 g which
can facilitate multiple brake modes including daily driving brake
and brake energy recuperation. Furthermore, the booster motor
in iBooster can automatically compensate pedal force and pro-
vide an adjustable pedal feel.
TRW released a decoupled electrohydraulic brake system in
[20]. A high-pressure accumulator with max 180 bar pressure,
coordinates with a bidirectional boost valve to control the fluid
flow directions, to increase or decrease wheel cylinder pressure
proportionally. Pedal simulator is adopted to provide force feed-
back to the driver. Proportional valves are used as both inlet and
C. Low-Cost outlet of wheel cylinders.
Electrified vehicles are more expensive than same class in- Solutions aforementioned can be generally classified into
ternal combustion engine based vehicles, due to added power two categories. One is high-pressure-accumulator-based, such
battery packs, related electronics, and control systems. Besides, as ECB and TRW’s configuration. The other category is
most sold production electrified vehicles are A00–A class ve- electric-booster based, like MK C1, iBooster and e-ACT. For
hicles or compact SUV, especially in developing countries like accumulator-based configurations, system imports costly linear
China, as shown in Table I [13]. These cars are extremely sen- solenoid valves, and a persistent high-pressure unit elevates the
sitive to parts cost, considering manufacturer’s suggested retail demand for machining precision and system seal. For electric-
price (MSRP), target end-users, and sales volume. Regenerative booster based configurations, complex master cylinder and mo-
brake-by-wire systems with fair performance, simple structure, tor transmission have increased the difficulty of manufacturing
and low-costing are more suitable under this circumstance. and assembling. Besides, the booster motor and integrated sen-
Several novel brake system configurations have been devel- sors used require complex control methods.
oped among automotive suppliers recently, which can partially Present authors have dedicated into regenerative braking sys-
meet aforementioned requirements for brake systems. tem development and applications under circumstance of Chi-
ADVICS and Toyota [2] presented a brake system configu- nese laggard manufacturing and automotive industry. Former
ration ECB, which can replace existing brake systems in both two generations of regenerative brake systems, i.e., the first ver-
ICEVs and EVs. Via the cooperation of newly developed mas- sion without pedal feel simulator [21], the second generation
ter cylinder, newly invented pressure applying module as well with only ABS function [22], have been successfully invented
as conventional pressure modulation unit, the proposed ECB and tested on commercial electrified vehicles. A complete frame
system can elevate regenerative efficiency and brake feel simul- of research work, including cooperative brake strategy, solenoid
taneously. The ECB system has been adopted on Toyota hybrid control technique, and methodology development of regenera-
vehicle i.e., Prius [14]. tive brake system, has been founded [23]–[26].

Authorized licensed use limited to: Air University. Downloaded on May 31,2021 at 07:48:16 UTC from IEEE Xplore. Restrictions apply.
11442 IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 67, NO. 12, DECEMBER 2018

Fig. 1. Schematic diagram of overall REHB system.

The new generation of regenerative electrohydraulic brake

system, which characterizes brake-by-wire functionality, i.e.,
REHB is released in this paper. In this generation, the architec-
ture of hydraulic system is simplified, but hydraulic control ac- Fig. 2. Schematic diagram of hydraulic pressure modulator.
curacy is improved significantly. High-frequency working noise
is eliminated. Brake pedal feel is refined greatly by importing
cooperative brake control logic, actuating valves and pumps
stroke simulator mechanism. Meanwhile, REHB has the capac-
and communicating with vehicle control unit (VCU). The
ity of integration of full vehicle dynamic control functions like
layout of hydraulic pressure modulator is depicted in Fig. 2,
ABS, ESC as well as ADAS like ACC and AEB. In addition, this
circled by the dashed rectangle.
generation of regenerative brake-by-wire system is the closest
The hydraulic pressure modulator is comprised of high-speed
to industrialization with the lowest cost.
valves, hydraulic pump, pressure sensors and pedal stroke sim-
The rest of the paper is organized as follows. System config-
ulators. Among 16 solenoid valves in total, eight ones (i.e., i-IV,
uration and function principles are presented in Section II. The
i-OV where i is LF, LR, RF and RR) are used for inlet and
design of brake force blending control strategy, which explains
outlet valves of four wheel-cylinders, resembling conventional
the apportionment of brake force and modulation approach of
four-channels hydraulic brake system. However, normally PWM
hydraulic pressure, is described detailedly in Section III. System
(Pulse Width Modulation) controlled inlet valves are replaced
modeling and simulation of typical brake scenarios are depicted
by four normally closed Current Amplitude Modulation (CAM)
in Section IV. Test bench setup as well as hardware-in-loop
method controlled valves. CAM method controlled valves are
bench experiments are provided in Section V followed by the
more precise in pressure modulation and have no working
concluding remarks in Section VI.
noise.
The six valves in upstream (ISV1 to ISV4, SV1, and SV 2
II. SYSTEM CONFIGURATION
in Fig. 2) are synchronously energized during normal brake
A. System Outline procedure. Once these six valves are engaged, the mechanical
The entire structure of the proposed by-wire brake system connection between upstream and downstream are cut off. The
REHB is sketched in Fig. 1. The brake pedal pressed by the brake fluid flows into stroke simulators from main cylinder when
driver is assisted by a vacuum booster to build up a high pressure brake pedal is depressed. Hence the master cylinder decouples
in master cylinder, whose value can be interpreted as the brake with wheel-cylinders by isolating all pressure fluctuations in
requirements of the driver. The master cylinder is connected to downstream, providing the driver a good pedal feedback with
hydraulic control unit (HCU), and then to calipers. HCU is used stroke simulators.
for modulating wheel cylinder pressures and providing pedal In the middle segment, a pump with six pistons coordinates
force feedback. It can cut off direct connection between master with a relief valve (RFV) that has a controllable crack pressure to
cylinder and wheel cylinders, fulfill by-wire brake actuation, provide a stabilized high pressure at inlets of all wheel-cylinders.
and imitate conventional pedal feel via stroke simulators. This high pressure is identical with pre-set crack pressure of
Hydraulic actuators are driven by the brake control board the relief valve. A normally closed solenoid valve is mounted
to modulate wheel cylinder pressures rapidly and precisely. The between outlet valves of all wheel cylinders and brake fluid
modulated pressures coordinate with regenerative brake force of tank serving as a failsafe valve, due to which, sufficient brake
electric motor to meet total brake demand. Hence, regenerative force can be guaranteed when a stuck outlet valve exists in the
braking is well realized. pipeline.

B. Hydraulic Control Unit C. Operation Modes

As the crucial component of proposed brake-by-wire system, Conventional hydraulic brake, regenerative brake as well as
HCU consists of brake control board and hydraulic pressure dynamic control functions such as ABS/ESC are integrated in
modulator. The brake control board is used for performing REHB.

Authorized licensed use limited to: Air University. Downloaded on May 31,2021 at 07:48:16 UTC from IEEE Xplore. Restrictions apply.
YUAN et al.: NOVEL REGENERATIVE ELECTROHYDRAULIC BRAKE SYSTEM: DEVELOPMENT AND HARDWARE-IN-LOOP TESTS 11443

Fig. 4. Force analysis of typical solenoid valve.

six solenoid valves in upstream are energized to lead the brake

fluid in master cylinder into stroke simulators and cut the con-
Fig. 3. Principle of collaborative regenerative brake. nections between upstream and downstream. The stroke simula-
tors provide a well-tuned pedal force feedback. Meanwhile, the
pump coordinates with valve RFV to build up a stabilized high
1) Conventional Hydraulic Brake: Under certain scenarios, hydraulic pressure before entries of all inlet valves. Further-
the drive motor of the vehicle cannot work as a generator, and more, the pressure in each wheel-cylinder is modulated quickly
then regenerative brake force cannot be implemented. For in- and precisely by corresponding inlet and outlet valves under
stance, the battery is full charged after a long distance downhill the control of designed pressure tracking algorithm. Four inlet
brake or the vehicle velocity is excessively low. Therefore, con- valves are controlled with CAM control method.
ventional hydraulic brake function is still indispensable. Though 3) ABS/ESC Function: ABS function will be activated when
all reasonable brake demand can be fulfilled by pump and pres- locking tendency of a certain wheel is detected during brake
sure modulation mechanism due to the brake-by-wire character procedure. Under ABS scenario, regenerative brake force evac-
of REHB, considering the energy consumption and system load, uates at a constant gradient to avoid jerks at vehicle level. All
under the working mode of conventional hydraulic brake, none six solenoid valves in upstream return to normal status and
of the components of REHB hydraulic pressure modulator is en- conventional ABS control architecture is resumed to take full
ergized. The connection status of the pipeline remains the same advantage of mature and verified ABS control algorithm. The
with conventional hydraulic brake system. The brake pedal is as- fluctuant pedal feel of conventional ABS is retained to alert the
sisted by vacuum booster. Unadjusted brake fluid flows directly driver to current road condition even though vibration isolation
from master cylinder into wheel cylinders, passing through iso- is attainable.
lating valves (i.e., ISV1-4) to build up demand pressure. Vehicle may lose stability when an emergent steering or turn-
2) Collaborative Regenerative Brake: Collaborative regen- ing on a slippery corner happens, where ESC function will be
erative brake, which is the core function of REHB, is fulfilled activated. All four isolation valves are energized isolating up-
by the cooperation of the drive motor that works as a genera- stream from downstream to avoid causing panic of the driver.
tor and the hydraulic brake system. The motor exerts a motor The pump-motor mechanism works at a high rotation speed
brake force on drive axle when used as a generator. The hy- to provide high-pressure fluid immediately. The inlet valves are
draulic brake system exerts friction brake forces on wheels. The regulated based on conventional ESC control logic to lead brake
total brake demand is met preferentially by motor brake force to fluid into certain wheel cylinder. Hence, single/double-wheel
maximize the efficiency of regenerative braking. If motor brake braking is fulfilled to generate a direct yaw torque to correct the
force is insufficient, hydraulic brake force will intervene to com- trajectory of the vehicle.
pensate the remaining part of brake demand. Thus, brake safety ABS and ESC functions driven by conventional algorithms,
is guaranteed and uniform brake feeling as conventional brake are omitted in this paper.
system is insured. The existence of regenerative brake force will
not be perceived. The transient maximum motor brake force is
determined by several factors including battery efficiency, vehi- D. Current Amplitude Modulation
cle velocity and ground adhesion etc. Therefore, the actual brake High-speed solenoid valves have been widely used in various
torque that the motor can exert on the driving axle keeps varying types of hydraulic pressure modulation mechanism. The general
during the whole brake procedure. Hence, the hydraulic pres- structure of a solenoid valve is illustrated in Fig. 4. The valve
sure in wheel cylinders needs to be modulated synchronously core with a spherical terminal is driven by electromagnetic force
and unremittingly according to the variation of actual motor Fe , spring force Fs , hydraulic force Fh , and supporting force
brake torque and total brake demand. FN , to regulate outlet pressure pou t from inlet pressure pin .
The work mechanism of collaborative regenerative brake is PWM control is the generally used valve control method to
depicted in Fig. 3. Under collaborative regenerative brake mode, adjust average electromagnetic force and track target pressure.

Authorized licensed use limited to: Air University. Downloaded on May 31,2021 at 07:48:16 UTC from IEEE Xplore. Restrictions apply.
11444 IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 67, NO. 12, DECEMBER 2018

The solenoid valve core vibrates between fully open and fully For hydraulic force Fh , we have,
closed positions during PWM working cycles to control the
Fh = Δp · Av (8)
gradient of pressure change. Subjected to the work principle
of PWM control and vibration of the valve core, the actual Av = πRv2 (cos α)2 (9)
pressure changes in the form of steps with glitches accompanied
with high frequency knocking noise. Considering the efficiency Δp = pin − pou t (10)
of regenerative brake, uniform driving feel and NVH (Noise- where Rv is spherical radius of valve terminal, and α is the
Vibration-Harshness) of the vehicle, increasing the accuracy of valve seat cone angle as shown in Fig. 4.
pressure modulation and eliminating working noise are of great Substituting (5), (7), (8), (9), and (10) into (2), then we have,
importance.
The pressure drop across a solenoid valve at ‘critical balanced KI ks (x0 + xm )
Δp = I (t) − (11)
state’ is linear with valve control current [27]. This relationship πRv (cos α)2
2 πRv2 (cos α)2
can be expressed as,
Let,
Δp = KI + Const (1) KI
K =
πRv (cos α)2
2
Where I is control current, Δp denotes pressure drop across
the valve, K is the current coefficient. The derivation of (1) is and,
expressed below. ks (x0 + xm )
Valve attached coordinate is set up as shown in Fig. 4. The Const = −
πRv2 (cos α)2
origin is the lowest point that valve core reaches. Critical bal-
anced state is the state where the valve core is equilibrated but Then we get equation (1).
about to move. It means that the valve core is just about to move According to (1), since pin equals to the crack pressure of
up to open the valve chamber or make hard contact with valve valve RFV Pr ef , which is known, if desired wheel cylinder pres-
seat to close the valve, i.e., FN = 0, and the displacement of sure is known, then the desired pressure drop Pdp dsr is known
valve core xv = 0. Then we have the equilibrium equation, too. So one can obtain open-loop feed-forward control current
easily from (1) [27]. Thus, wheel pressure can be regulated
−Fe + Fs + Fh = 0 (2) precisely via modulating coil current, i.e., Current Amplitude
Modulation (CAM).
Electromagnetic force is expressed as, CAM control method adjusts the current amplitude exerted
on solenoid coil based on pressure drop requirement, which is
(N I)2 the pressure difference between pump outlet and wheel demands
Fe = (3)
2Rg l in REHB case. The valve core only moves once between two
terminal positions during one modulation cycle. Therefore, there
Here I and N are coil current and number of windings re- is no high frequency working noise.
spectively. l and Rg are the length and magnetic reluctance of However, the relationship expressed above is founded based
air gap. (3) can be linearized at critical balanced state as, on assumptions of that the valve works at an ideal position and
 
∂Fe  ∂Fe  hydrodynamic force is ignored. Whereas, the actual relationship
Fe = · I (t) + · xv (t)
∂I I =I (t) ∂xv  x v =x v (t)
is not rigidly linear. Open-loop feed-forward control like [27] is
too roughly. To improve the tracking performance and robust-
= KI I (t) + Kx v xv (t) (4) ness to external disturbances, a closed-loop control algorithm
based on CAM, for wheel pressure modulation is developed.
KI and Kx v are first-order current and displacement This novel control method adopts a feed-forward lookup table
coefficients. xv = 0 is satisfied at critical balanced point, then and a simple feed-back PID controller, which will be introduced
(4) can be simplified as, more minutely in following sections.
Although, CAM can be used for both inlet and outlet, in
Fe = KI I (t) (5) REHB, only inlet valves are controlled with CAM method. Out-
let valves are still controlled with PWM method, since pressure
Spring force Fs can be derived as follow, decreasing demand can be easily achieved in only several duties
with low frequency due to stable atmospheric pressure at exit of
Fs = ks (x0 + xm − xv ) (6)
outlet valves. Therefore, there is no need for a dedicated CAM
where x0 is the pre-tension spring length, xm is the maxi- chip for outlet valve which will increase system cost in turn.
mum displacement of spring, and ks is the stiffness of return
spring. xv equals to 0 at critical balanced point, then (6) can be III. REGENERATIVE BRAKING CONTROL STRATEGY
simplified as, The control strategy of REHB system integrates several mod-
ules, including regenerative braking control module, ABS con-
Fs = ks (x0 + xm ) (7) trol module, ESC control module, and fault diagnose module

Authorized licensed use limited to: Air University. Downloaded on May 31,2021 at 07:48:16 UTC from IEEE Xplore. Restrictions apply.
YUAN et al.: NOVEL REGENERATIVE ELECTROHYDRAULIC BRAKE SYSTEM: DEVELOPMENT AND HARDWARE-IN-LOOP TESTS 11445

Fig. 6. Brake demand distribution strategy for all driven type.

Fig. 7. The structure of close-loop PWM CAM coupling control.

Fig. 5. Architecture of regenerative braking control strategy.

etc. A central unit is used to deploy tasks between different Brake demand distribution includes two tasks, i.e., division
modules based on current scenarios. As mentioned in the pre- between vehicle axles as well as apportionment between regen-
vious section, ABS and ESC functionalities remain the same erative and friction brakes.
with regular ones, therefore only regenerative braking control On conventional passenger vehicles, the relationship between
strategy will be introduced in this section. The overall control front and rear brake forces is proportional to each other with
structure of regenerative braking control strategy is a tri-level fixed ratio. In order to take full advantage of conventional mature
close-loop control architecture, as depicted in Fig. 5. brake system experience, this proportional relationship has been
At the top level, Brake Demand Distribution is in charge of kept as shown in Fig. 6 denoted with line β.
apportioning brake demand between motor brake and hydraulic The apportionment between regenerative and hydraulic brake
brake from total brake demand Tdem an d at both cooperative forces obeys the principle of maximizing motor brake force. The
regenerative brake stage and regenerative brake exit stage. The brake demand on drive-axles (front or rear) is preferentially met
outputs of top level are target motor brake torques Mtr q tg t and by motor brake force. If the brake demand allocated exceeds
target hydraulic pressures Ptg t . Vehicle current status Vin f o , the mechanical limit of the motor along with the augment of
current motor torque Tact , and current hydraulic pressure info brake demand, hydraulic brake will intervene to supplement the
Pact are main input arguments. insufficient part.
At the middle level, Pressure Modulation Algorithm module The blending of regenerative and hydraulic forces equals to
is set up, which consists a Close-loop PWM CAM Coupling total brake demand throughout. The driver will not be aware
Control sub-module and a Logic Threshold Control sub- of the existence of regenerative braking force, friction braking
module. Ptg t from upper level and real-time hydraulic pressure force or switching between them. Hence, a uniform brake feel in
info Pact are inputted into Pressure Modulation Algorithm vehicle level can be guaranteed, and a good pedal force feedback
module to calculate reference control commands (INr ef , can be provided due to the stroke simulators.
OU Tr ef ) for inlet valve and outlet valves. Reference com-
mands are tuned by Logic Threshold Control, and Hydraulic B. Close-Loop PWM CAM Coupling Control
Control Unit drive signal HCU drv are outputted.
At the bottom level, Mtr q tg t is retransmitted to motor by This module is a close-loop wheel cylinder pressure con-
VCU via CAN bus. Desired motor brake force is performed trol module. Control inputs for a single channel are calculated.
immediately. HCU drv drives the hydraulic system to achieve The structure of Close-loop PWM CAM Coupling Control for
desired hydraulic pressure. Thus, motor brake force coordinates hydraulic pressure modulation is shown in Fig. 7. To improve
with hydraulic brake force, to meet total brake force demand. the robustness and precision of the control algorithm, a feed-
forward and feed-back scheme is employed for both inlet and
outlet valves.
A. Brake Demand Distribution
For inlet valves, a current map derived from the linear char-
Based on the HCU depicted in Fig. 2, four channels have acteristics as equation (1) is set up based on bench test exper-
the same structure and can perform identical functions. Hence, iments. A simple pressure feed-back PID controller is utilized
REHB system suits all driven types including front/rear-drive to eliminate external disturbances. The reference inlet control
vehicles and all-wheel-drive vehicles. input is outputted. Resembling with inlet valves, a PWM map

Authorized licensed use limited to: Air University. Downloaded on May 31,2021 at 07:48:16 UTC from IEEE Xplore. Restrictions apply.
11446 IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 67, NO. 12, DECEMBER 2018

Fig. 9. Eight DOFs vehicle dynamic model.

in AMESim. MATLAB/Simulink and AMESim can perform

co-simulation via dedicated software communication interface.
Fig. 8. State machine of logic threshold control.
A. Vehicle Dynamics
and PID controller are set up for outlet valves. Two integrated A three-degree planar motion of vehicle body dynamics is
lookup tables provide open loop control input based on real-time modeled in vehicle body reference coordinate OXY Z coordi-
difference between Ptg t and Pact , i.e., Per r . The parameters nate as shown in Fig. 9. Three DOFs (degree of freedom) are
of PID controllers are semi-empirical and optimized based on body motion along OX and OY axles and the yaw rate around
Hardware-In-Loop tests. OZ axle.
Equations of motion in longitudinal, lateral, and yaw direc-
C. Logic Threshold Control tions can be expressed as follows [23].
Reference control commands calculated cannot be used to ⎧
⎨ V̇x = Vy γ + m Fx
⎪ 1
drive pumps and valves in HCU directly, because un-tuned com-
V̇y = −Vx γ + m1 Fy (12)
mand may cause system instability. Logic Threshold Control ⎪
⎩ γ̇ = 1 M
module has the structure of three-node state machine. It can Jz z
tune commands and switch working states between three dif-
ferent work states, i.e., pressure hold, increase and decrease as Here Vx and Vy denote the longitudinal and lateral speeds
shown in Fig. 8. of vehicle body. γ is the yaw rate around OZ axle. m is the
Two grades of increasing and decreasing are set up i.e., vehicle mass, Jz is the yaw inertia. Fx , Fy and Mz are to-
M oderate and F ast. The transfer conditions between dif- tal forces/moment generated by tire-road forces in direction of
ferent states are regulated in the form of multiple thresholds OX, OY and around OZ respectively, and are defined in (13)
(threshold1 − threshold8 ) based on the pressure residual er- below.
⎧
ror Per r . These thresholds are tuned based on bench tests. The ⎪
⎪ Fx = (Fx1 + Fx2 ) cos δ − (Fy 1 + Fy 2 ) sin δ
output signals of this module, including actual inlet command ⎪
⎪
⎪
⎪ + Fx3 + Fx4
Inlet cmd, outlet command Outlet cmd, and pump command ⎪
⎪
⎪
⎪
P ump cmd, are amplified by triodes on control board and then ⎨Fy = (Fx1 + Fx2 ) sin δ + (Fy 1 + Fy 2 ) cos δ
⎪
used to drive HCU. + Fy 3 + Fy 4 (13)
⎪
⎪
⎪ Bf
⎪Mz = 2 [(Fx2 − Fx1 ) cos δ + (Fy 1 − Fy 2 ) sin δ]
⎪
⎪
D. Reg Exit Control ⎪
⎪
⎪
⎪ + B2r (Fx3 − Fx4 ) + a [(Fx1 + Fx2 ) sin δ
⎪
⎩
Brake performance deteriorates when motor speed is lower + (Fy 1 + Fy 2 ) cos δ] − b (Fy 3 + Fy 4 )
than a certain threshold. Therefore, regenerative bake needs to
exit at the end stage of brake procedure. In REHB, at the exiting In equations above, Fxi is the tangential tire-ground force
stage, motor brake decreases at a fixed speed and hydraulic of wheel i, and Fy j represents the side tire-ground force of
brake increases according to brake demand simultaneously. wheel j. a, b are the distances between COG and front/rear axles.
Bf and Br are the lengths of front and rear wheel tracks. Air
IV. MODELING AND SIMULATION resistance is neglected.
In order to verify the reasonability of HCU architecture and
testify the control performance of the proposed control strategy B. Wheel and Tire
preliminarily, a compact electric vehicle with front-drive motor Wheel model is the rotational dynamics model of wheel. Tire
and developed REHB system are modeled. The models of model is tire-ground force model.
vehicle body dynamics, wheel-tire assembly, motor, battery 1) Wheel Rotational Dynamics Model: A schematic force
and controllers were built in MATLAB/Simulink. The model analysis of wheel i is depicted in Fig. 10. Fz i is the normal force
of HCU hydraulic part including pumps and valves is built exerted on wheel i. Vω i is the generalized velocity of wheel i.

Authorized licensed use limited to: Air University. Downloaded on May 31,2021 at 07:48:16 UTC from IEEE Xplore. Restrictions apply.
YUAN et al.: NOVEL REGENERATIVE ELECTROHYDRAULIC BRAKE SYSTEM: DEVELOPMENT AND HARDWARE-IN-LOOP TESTS 11447

Fig. 12. CAM controlled valve in AMESim.

teristic time. However, compared with other actuators, electric

motor is highly faster in response to control command. In order
Fig. 10. Schematic diagram of wheel dynamic model.
to simplify the model construction with the premise of enough
precision, a first-order lag plant system with same characteristic
time 0.02 s is adopted as the model of drive motor. It can be
expressed with transfer function as follows;
1
G (s) = (16)
0.02s + 1

D. Battery
The battery model is modeled based on charging/discharging
efficiency - SOC map provided by Chery Motor Company [29].
It can be used to monitor the variety of bus voltage and current,
which reflects the power of motor. The battery capacity is set to
45 Ah and the nominal voltage is 336 V.

Fig. 11. Motor-dynamometer bench and step response test result.

E. Hydraulic Pressure Modulator
Hydraulic pressure modulator, including pumps, valves,
Tdi and Tbi are drive torque and brake torque exerted on wheel cylinders, and calipers, is built in a commercial software
i respectively. For driven wheels, Fdi ≡ 0. AMESim via a parametric and modularized method. All pro-
According to torque equilibrium, the wheel rotational dynam- totype modules from AMESim library has been testified by
ics of a front-drive vehicle can be derived as followed. engineering projects. Pedal feel is not concerned in simulations.
 Therefore, pedal mechanism, master cylinder, and stroke simu-
Iw i ωi = Tdi − Tbi − Fxi r (i = 1, 2, 3, 4) lator have been omitted from the system.
(14)
Tdi = 0 (i = 3, 4) CAM controlled valve is modeled based on solenoid valve
work mechanisms as shown in Fig. 12. The magnetic force was
Here ωi is wheel rotational speed, and r is tires effective refined by considering the dynamics of valve rod to improve
rolling radius. Iw i is the inertia of wheel-tire assembly i. the fidelity of the model. The parameters as shown in Table III
2) Tire-Road Force Model: In this paper, Pacejka’s Magic within Appendix, are provided by an Chinese hydraulic brake
Formula method [28] is adopted to model tire-road force. manufacturing company AP G CO., LT D [30].
Under single condition, both longitudinal and transverse After choosing appropriate prototype models and setting rea-
forces can be expressed with a uniform form as follows: sonable parameters, all components are connected with pipeline


Y (x) = D sin C tan−1 Bϕ (1 − E) + E tan−1 (Bϕ) as shown in Fig. 13.
+ Sv (15)
F. Simulations
Here D, C, B, E, ϕ, and Sv are all factors calculated by em-
In order to verify the reasonability of HCU architecture and
pirical equations based on bench test results. Y (x) can be seen
testify effectiveness of proposed control strategy, simulations
as either longitudinal or transverse tire-road forces.
are conducted under typical one-step braking scenarios.
One-step braking is defined as follow. The initial velocity of
C. Electric Motor target front-drive electric vehicle is 75 km/h. The brake demand
Based on the motor step-response bench test result of a motor- raises from zero with a fix slope, and maintains at maximum
dynamometer bench as shown in Fig. 11, the motor is a typical brake demand to the end of whole brake procedure. Three
second-order control system with 3T = 0.06 s, here T is charac- different brake demand levels are adopted as shown in Fig. 14.

Authorized licensed use limited to: Air University. Downloaded on May 31,2021 at 07:48:16 UTC from IEEE Xplore. Restrictions apply.
11448 IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 67, NO. 12, DECEMBER 2018

demand. The maximum available motor brake force increases

along with the decrease of motor rotational speed. However,
brake demand reaches its maximum value. In order to exhaust
regenerative brake force capacity, hydraulic brake force is
moderated and decreases according to the difference between
total brake demand and motor actual brake force. Hence,
regenerative brake force is maximized. The motor enters the
zone of constant torque at the end of stage S2.
At the stage of S3, the brake demand and motor brake force
maintain at respective maximum valves and the hydraulic pres-
sure keeps at a constant valve as well. Stage S3 ends when motor
rational speed is lower than threshold set (550 rpm in this case),
under which, the motor performance deteriorates.
At the stage of S4, motor brake force deteriorates. Hydraulic
brake force increases along with the exit of motor brake force
that decreases at a fixed slope to avoid potential jerks on the
vehicle. S4 continues to the end of brake procedure.
According to simulation results, motor brake force works with
hydraulic brake force cooperatively. The addition of two kinds
of brake force meets brake demand at all circumstances, which
Fig. 13. Overall hydraulic system model in AMESim. has testified the reasonability of proposed HCU architecture and
control strategy.

V. HARDWARE-IN-LOOP (HIL) BENCH TEST

A prototype of HCU and a split control board are manufac-
tured by AP G CO., LT D. A HIL test bench was set up to test
the performance of actual REHB system as shown in Fig. 16.
Compared with models used in simulations, the HCU, hydraulic
pipeline and brake drums/calipers are replaced with actual ob-
jects. Vehicle dynamics, motor and battery models are converted
Fig. 14. Simulation scenarios set up – three demand levels. into binary code and downloaded to dSPACE Real-time simu-
lation platform via MATLAB. The brake control unit (BCU)
For mild scenario, brake demand, which is measured with and dSPACE exchange data via CAN bus Network. A host PC
pressure in master cylinder, increases by 2 MPa/s to 3 MPa in carrying a software named Controldesk NG is used to acquire
1.5 seconds. For moderate scenario, brake demand increases by data and control process of bench tests.
5 MPa/s to 3 MPa in 0.6 seconds. For severe scenario, brake The relationship between drive commands and pressure drop
demand increases by 5 MPa/s to 4.5 MPa in 0.9 seconds. These of CAM controlled valves was measured and mapped firstly
brake demand sets can cover all normal straight-line brake. to initialize inlet valve control. Then, identical one-step brake
The performance of REHB system and control effect of pro- scenarios with simulations were repeated on the test bench.
posed control strategy based on simulation results are depicted More aspects of performance are concerned, including the per-
in Fig. 15 contrastively. formance of hydraulic brake on driven axle and the regenerative
The brake demand, motor brake force, hydraulic brake force brake efficiency.
on drive axle are all scaled with brake pressure. Curve of vehicle
velocity is also attached to show the status of target vehicle.
There are four stages observed during a typical one-step brake A. CAM Control Test
procedure according to simulation result in Fig. 13. A SC900717 chip produced by Freescale [31] that can per-
At the stage of S1, brake demand is met by motor brake force form current modulation is integrated into BCU. CAM control
only on drive-axle, before motor brake force reaches the physical test is conducted as follow. A pre-set sequence of current control
torque limit corresponding to current rotational speed. Hence, commands is exerted on inlet valve of HCU. The pressure drop
the velocities of vehicle at the end of stage S1 are the same in sequence across the inlet valve corresponding to current control
three situations, whereas the durations of S1 under moderate command sequence is recorded. The test result is illustrated in
and severe scenarios are shorter due to higher brake force. Fig. 17.
At the stage of S2, motor brake force is not sufficient and The least squares fitting of sequences of current control
enters the constant power zone of motor. Motor brake torque command and pressure drop exposes the approximate linear
is inversely proportional to its rotational speed. Hydraulic relationship between them. The data acquired was mapped into
brake force is exerted by the HCU to compensate remain brake a current map in Close-loop PWM CAM Coupling Control

Authorized licensed use limited to: Air University. Downloaded on May 31,2021 at 07:48:16 UTC from IEEE Xplore. Restrictions apply.
YUAN et al.: NOVEL REGENERATIVE ELECTROHYDRAULIC BRAKE SYSTEM: DEVELOPMENT AND HARDWARE-IN-LOOP TESTS 11449

Fig. 15. Typical one-step brake simulation results (Front axle).

For stage S2, motor brake is insufficient, so hydraulic brake

is exerted to supplement brake demand. Stage S2 ends when
motor work mode changes to constant torque zone.
For stage S3, motor outputs its maximum brake torque, and
hydraulic system stays in pressure hold state.
For stage S4, motor brake exits with a fixed ratio, and hy-
draulic brake force increases to compensate insufficient part of
brake demand. No jerks are observed on vehicle level.
According to bench results, motor brake force works with
hydraulic brake force cooperatively. The equivalent brake force,
which is the addition of two kinds of brake force, meets brake
demand at all circumstances, The HIL test results have proven
the cooperative work of motor and hydraulic system and the
effectiveness of proposed brake demand distribution algorithm.
Besides, no high frequency work noise was observed during
bench tests.
The results on driven axle are also monitored to observe the
Fig. 16. Test bench set up based on real hydraulic system hardware.
holistic performance of hydraulic pressure modulation as shown
in Fig. 19.
The hydraulic pressure increasing on rear axle has a slug-
gish lag compared with brake force increase on front axle ac-
cording to the test result, due to the fact that the response of
hydraulic system is much slower than motor. According to the
result curves, hydraulic pressure tracks brake demand on rear
axle well similarly, proving the effectiveness of hydraulic pres-
sure modulation.
Regenerative brake efficiency is adopted to evaluate the en-
ergy recovering capability of REHB system. Regenerative brake
Fig. 17. Bench test result of current modulation. efficiency in one-step brake is defined as follow,
⎧
⎨ ρ = E a v a i l a b l e × 100%
Er e c o v e r e d
module which has been expressed in section Regenerative ⎪
braking control strategy. Er ecov er ed = − ∫ Vb × Cb dt (17)
⎪
⎩
Eav ailable = 12 mv02 − ∫ mgfg vdt − ∫ CD Av 3 dt
B. One-Step Brake Bench Test
Brake demands are set exactly identically with simulation Er ecov er ed is the energy recovered by REHB and charged into
scenarios. Brake demand, motor brake force, hydraulic pressure the battery. Eav ailable is the total kinetic energy exclusive of en-
and equivalent brake force are all evaluated with the scale of ergy consumed by wind resistance and rolling resistance which
pressure as shown in Fig. 18 for all three scenarios. can be regarded as total energy that is ready to be recovered. Vb
Identical four stages and similar trend of curve variation with and Cb are battery bus voltage and current. Cb is negative when
simulation results can be observed in test results. the battery is being charged. v0 is the initial velocity of target
For stage S1, total brake demand can be satisfied by mo- vehicle. g is the gravitational constant and is set as 9.8. fg is
tor only, before brake demand reaches constant power zone of the friction coefficient. Data sets acquired from CAN Bus con-
motor. taining battery voltage and current as well as vehicle velocity

Authorized licensed use limited to: Air University. Downloaded on May 31,2021 at 07:48:16 UTC from IEEE Xplore. Restrictions apply.
11450 IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 67, NO. 12, DECEMBER 2018

Fig. 18. Typical one-step brake bench test results (Front axle).

Fig. 19. Typical one-step brake bench test results (Rear axle).

Fig. 20. Typical one-step brake bench test results (Rear axle).

are processed and results are shown in Fig. 20. The regenerative C. Comparisons and Improvements
brake efficiency can be calculated as follow. 1) Precision Improvement: One way to evaluate the perfor-
For mild brake demand scenario,
mance of hydraulic modulation precision and blending control
140.32 strategy is to calculate the jerk on the vehicle level during brake.
ρ1 = × 100% = 46.32%
302.93 Jerk is the second derivative of vehicle velocity, which can re-
flect the gradient of total brake force. The HIL jerk data of
For moderate brake demand scenario,
previous generation of regenerative brake system EABS [32],
134.16 as well as the jerk data of REHB was depicted in the same figure
ρ2 = × 100% = 43.93%
305.19 as shown in Fig. 21.
For severe brake demand scenario, Jerk usually happens when brake starts and when regenera-
tive brake exits. The maximum absolute jerk of EABS is about
98.27 4 m/s3 , while for REHB case, the maximum absolute jerk is
ρ3 = × 100% = 31.88%
308.26 about 2 m/s3 , especially for regenerative brake exiting stage, the
Regenerative brake efficiency decreases along with the inten- jerk has been whittled distinctly.
sifying of brake demand because resistance consumed energy 2) Comprehensive Assessment: In this segment, more as-
are decreasing, the brake distance are shorter, and hydraulic sessment criteria are addressed among EABS, REHB, and the
brake undertakes more responsibility along with intensifying of most advanced brake-by-wire system iBooster + ESP from
brake demand. Bosch. iBooster is an electric motor based booster, but cannot

Authorized licensed use limited to: Air University. Downloaded on May 31,2021 at 07:48:16 UTC from IEEE Xplore. Restrictions apply.
YUAN et al.: NOVEL REGENERATIVE ELECTROHYDRAULIC BRAKE SYSTEM: DEVELOPMENT AND HARDWARE-IN-LOOP TESTS 11451

one-step-brake simulations were conducted. Simulation results

showed the reasonability of proposed structure and effectiveness
of designed control strategy.
A prototype of REHB was manufactured by an outsourced
company. Regenerative brake test bench was set up based on
the prototype product. CAM control test and identical one-step-
brake experiments were conducted respectively. The relation-
ship of current and pressure drop was measured and mapped
into lookup tables for underlying control. The results of one-
step-brake including drive and driven axle were analyzed to
evaluate holistic performance of REHB. The energy recovered
in bunch of tests was also calculated. The regenerative effi-
ciencies were 46.32%, 43.93% and 31.88% of total recoverable
Fig. 21. Test bench set up based on real hydraulic system hardware. energy respectively in three bench tests.
Comparison with precious generation EABS and iBooster
perform pressure modulation. ESP are used to modulate wheel showed that, REHB which has very simple structure, can im-
pressures like HCU in REHB. Main criteria are regenerative prove the performance of regenerative brake-by-wire system in
efficiency, modulation accuracy, functions, pedal feel, fail safe, all scales with expressly low cost.
initiative braking, and most important, system cost.
For the same class vehicles, EABS and REHB can recuper- APPENDIX
ate the same energy during brake. But there is no A0 class
TABLE III
vehicle equipped with iBooster and ESP that is reported. The STRUCTURAL PARAMETERS USED FOR VALVE MODELING
modulation accuracy of EABS is only 0.3 MPa, while REHB
and iBooster + ESP can achieve astonishing 0.2 MPa [33]. Only
ABS function and regenerative braking (RBS) can be performed
on EABS, while ADAS function including ACC and AEB can
carry out on REHB and iBooster + ESP. That is because the
latter two systems are capable of executing initiative braking.
There is no pedal stroke simulator on EABS, so its pedal feel is
much worse than REHB and iBooster + ESP. Degraded brake
performance is shown on both EABS and iBooster + ESP, be-
cause no booster exists when the system fails. But in REHB, REFERENCES
vacuum booster can still assist the driver to build up un-degraded [1] K. Bimbraw, “Autonomous cars: Past, present and future a review of the
brake pressure even when system fails. Last, an iBooster charges developments in the last century, the present scenario and the expected
more than $1000 [34], but EABS and REHB are highly cheaper future of autonomous vehicle technology,” in Proc IEEE 12th Int. Conf.
Inform. Control Automat. Robot., 2015, vol. 1, pp. 191–198.
with the price of about $50. More explicit comparison is listed [2] Y. Masuda, Y. Yamasoe, Y. Kuki, T. Okano, and K. Uchida, “Development
in Table II. of an electronically controlled brake system for fuel-efficient vehicles,” in
Proc. SAE World Congr. Exhib., Detroit, MI, USA, Apr. 12–16, 2016.
[3] S. A. Oleksowicz et al., “Regenerative braking strategies, vehicle safety
VI. CONCLUSION and stability control systems: Critical use-case proposals,” Vehicle Syst.
Dyn., vol. 51, no. 5, pp. 684–699, 2013.
This paper has proposed a novel solution of by-wire brake [4] E. V. Bunyaeva, V. G. Skorik, S. V. Vlasevskii, and D. S. Fokin, “A method
system REHB that has considered the requirements of the re- for improving the energy efficiency of an alternating current electric lo-
comotive in the regenerative braking mode,” Russian Elect. Eng., vol. 87,
generative braking function at structure development stage. The no. 2, pp. 73–76, 2016.
control strategy, including brake demand distribution and pres- [5] Z. Liu, W. J. Ortmann, B. Nefcy, D. Colvin, and F. Connolly, “Methods
sure modulation algorithm, was well designed to coordinate of measuring regenerative braking efficiency in a test cycle,” SAE Int. J.
Alternative Powertrains, vol. 6, no. 1, pp. 103–112, 2017.
motor brake force and hydraulic brake force. Brake demand [6] C. S. N. Kumar and S. C. Subramanian, “Cooperative control of re-
distribution was design to follow the principle of exhausting the generative braking and friction braking for a hybrid electric vehicle,”
capacity of the motor to improve regenerative brake efficiency. Proc. Institution Mech. Eng. Part D, J. Automobile Eng., vol. 230, no. 1,
pp. 103–116, 2016.
The exit of regenerative braking process was also optimized [7] F. Naseri, E. Farjah, and T. Ghanbari, “An efficient regenerative braking
to avoid potential jerks on vehicle. In the aspect of underly- system based on battery/supercapacitor for electric, hybrid, and plug-in
ing control, Current Amplitude Modulation control method was hybrid electric vehicles with BLDC motor,” IEEE Trans. Veh. Technol.,
vol. 66, no. 5, pp. 3724–3738, May 2017.
adopted to improve the accuracy of hydraulic pressure modula- [8] B. Eichhorn, S. König, and T. Ullrich, “Electronic brake control for greater
tion and eliminate valve vibration noise. active safety,” Auto Tech Rev., vol. 4, no. 4, pp. 30–33, 2015.
High fidelity models of vehicle dynamics, motor, battery, hy- [9] S. Ko, C. Song, and H. Kim, “Cooperative control of the motor and
the electric booster brake to improve the stability of an in-wheel elec-
draulic system and controllers were built in MATLAB/Simulink tric vehicle,” Int. J. Automot. Technol., vol. 17, no. 3, pp. 447–456,
and AMESim co-simulation platform. Three levels of 2016.

Authorized licensed use limited to: Air University. Downloaded on May 31,2021 at 07:48:16 UTC from IEEE Xplore. Restrictions apply.
11452 IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 67, NO. 12, DECEMBER 2018

[10] J. Ko, S. Ko, H. Son, B. Yoo, J. Cheon, and H. Kim, “Development of [33] J. Becker, M.-B. A. Colas, S. Nordbruch, and M. Fausten, “Bosch’s vision
brake system and regenerative braking cooperative control algorithm for and roadmap toward fully autonomous driving,” in Road Vehicle Automa-
automatic-transmission-based hybrid electric vehicles,” IEEE Trans. Veh. tion, Berlin, Germany: Springer, 2014, pp. 49–59.
Technol., vol. 64, no. 2, pp. 431–440, Feb. 2015. [34] Ebay, 2016–2018 Tesla Model S 75D iBooster master cylinder electric
[11] C. F. Lee and C. Manzie, “Active brake judder attenuation using an elec- assembly OEM. [Online]. Available: https://www.ebay.com/itm/2016
tromechanical brake-by-wire system,” IEEE/ASME Trans. Mechatronics, -2018-TESLA-MODEL-S-75D-iBOOSTER-MASTER-CYLINDER-
vol. 21, no. 6, pp. 2964–2976, Dec. 2016. ELECTRIC-ASSEMBLY-OEM/292536541481?hash=item441c896129:
[12] L. Yu, X. Liu, Z. Xie, and Y. Chen, “Review of brake-by-wire system g:YXkAAOSw7Bda31zP
used in modern passenger car,” in Proc. ASME Int. Des. Eng. Tech. Conf.
Comput. Inf. Eng. Conf., 2016, pp. V003T01A020.
[13] Zach, 2017 Electric car sales — USA, China, and Europe (month by
month). 2018. [Online]. Available: https://evobsession.com/2017-electric-
car-sales-us-china-europe-month-month/ Ye Yuan was born in Jilin, China in 1991. He re-
[14] H. Watanabe, T. Segawa, T. Okuhira, H. Mima, and N. Hoshikawa, “A ceived the B.S. degree in automotive engineering
custom integrated circuit with on-chip current-to-digital converters for from the Beijing Institute of Technology, Beijing,
active hydraulic brake system,” SAE Int. J. Passenger Cars, Electron. China, in 2013. He is currently working toward the
Elect. Syst., vol. 9, no. 1, pp. 188–195, 2016. Ph.D. degree with the Department of Automotive En-
[15] T. Oshima, N. Fujiki, S. Nakao, T. Kimura, Y. Ohtani, and K. Ueno, gineering, Tsinghua University, Beijing, China. His
“Development of an electrically driven intelligent brake system,” SAE Int. research interests include regenerative braking, fault
J. Passenger Cars – Mech. Syst., vol. 4, no. 1, pp. 399–405, 2011. tolerant control, and optimization control.
[16] J. Zhang, Y. Yuan, C. Lv, and Y. Li, “Modeling and analysis of regenerative
braking system for electric vehicle based on AMESim,” in Proc. IEEE Int.
Conf. Mechatronics Automat., 2015, pp. 1307–1312.
[17] K. Meuer, “New requirements and solutions in brake development–new
challenges for brake systems in prototype vehicles targeting highly au-
tomated driving,” in Proc. 7th Int. Munich Chassis Symp. 2016, 2017,
pp. 761–776.
[18] C. Riese, A. Verhagen, S. Schroeter, and F. Gauterin, “Comparison of
a state of the art hydraulic brake system with a decentralized hydraulic
Junzhi Zhang received the B.S., M.S., and Ph.D. de-
brake system concept for electric vehicles,” presented at the SAE Technical grees from Jilin University, Changchun, China. He
Paper Series, Warrendale, PA, USA, 2017.
is currently a Professor with the State Key Labo-
[19] C. Shahan, Bosch Takes Tesla Model S Automated Driving Tests,
ratory of Automotive Safety and Energy, Tsinghua
2015. [Online]. Available: https://cleantechnica.com/2015/06/17/bosch-
University, China. His research interests include re-
takes-tesla-model-s-automated-driving-tests/ generative braking, hybrid vehicle powertrain, energy
[20] D. Savitski et al., “The new paradigm of an anti-lock braking system
management, autonomous vehicle test, and load em-
for a full electric vehicle: experimental investigation and benchmarking,”
ulator design.
Proc. Institution. Mech. Eng. Part D, J. Automobile Eng., vol. 230, no. 10,
pp. 1364–1377, 2016.
[21] J. Zhang, C. Lv, J. Gou, and D. Kong, “Cooperative control of regen-
erative braking and hydraulic braking of an electrified passenger car,”
Proc. Institution. Mech. Eng. Part D, J. Automobile Eng., vol. 226, no. 10,
pp. 1289–1302, 2012.
[22] J. Zhang, C. Lv, X. Yue, M. Qiu, J. Gou, and C. He, “Development of the
electrically-controlled regenerative braking system for electrified passen-
ger vehicle,” presented at the SAE Technical Paper Series, Warrendale,
PA, USA, 2013. Yutong Li received the B. Eng. degree from Jilin
[23] Y. Yuan, J. Zhang, C. Lv, and Y. Li, “Design and performance analysis University, Changchun, China, in 2012, and the Ph.D.
of a novel regenerative braking system for electrified passenger vehicles,” degree from Tsinghua University, Beijing, China, in
SAE Int. J. Mater. Manuf., vol. 9, no. 3, pp. 699–706, 2016. 2017. He is currently working toward the Postdoc
[24] C. Lv, J. Zhang, Y. Li, D. Sun, and Y. Yuan, “Hardware-in-the-loop sim- with the Department of Transportation, Tongji Uni-
ulation of pressure-difference-limiting modulation of the hydraulic brake versity, Shanghai, China. His research interests in-
for regenerative braking control of electric vehicles,” Proc. Institution. clude topics in decision making and control under
Mech. Eng. Part D, J. Automobile Eng., vol. 228, no. 6, pp. 649–662, uncertainties (application to autonomous vehicles),
2014. simulation based verification and validation, and
[25] C. Lv, J. Zhang, Y. Li, and Y. Yuan, “Novel control algorithm of braking powertrain architecture optimization for performance
energy regeneration system for an electric vehicle during safety-critical and fuel economy.
driving maneuvers,” Energy Convers. Manage., vol. 106, pp. 520–529,
2015.
[26] Y. Yuan, J. Zhang, Y. Li, and C. Lv, “Regenerative brake-by-wire sys-
tem development and hardware-in-loop test for autonomous electrified
vehicle,” in Proc. SAE World Congr. Experience, Detroit, MI, USA, Apr.
4–6, 2017.
[27] J. Zhang, C. Lv, X. Yue, Y. Li, and Y. Yuan, “Study on a linear relationship
between limited pressure difference and coil current of on/off valve and
its influential factors,” ISA Trans., vol. 53, no. 1, pp. 150–161, Jan. 2014.
Chao Li received the B.S. degree from Jilin Uni-
[28] H. B. Pacejka and E. Bakker, “The magic formula tyre model,” Vehicle
Syst. Dyn., vol. 21, no. S1, pp. 1–18, 1992. versity, Changchun, China. He is currently working
toward the Ph.D. degree with the State Key Labo-
[29] T. Jeng, “Chery,” En.wikipedia.org, 2018. [Online]. Available: https://
ratory of Automotive Safety and Energy, Tsinghua
en.wikipedia.org/wiki/Chery
University, Beijing, China. His research interests in-
[30] “About Us,” Zhejiang Asia-pacific Mechanical & Electronic Co., Ltd,
2018. [Online]. Available: http://www.chinaapg.com/about.html clude regenerative braking, nonlinear control, vehicle
system dynamics, and energy management.
[31] Freescale, SC900717 User Manual, 2014.
[32] Z. Junzhi, L. Yutong, L. Chen, and Y. Ye, “New regenerative braking
control strategy for rear-driven electrified minivans,” Energy Convers.
Manage., vol. 82, pp. 135–145, 2014.

Authorized licensed use limited to: Air University. Downloaded on May 31,2021 at 07:48:16 UTC from IEEE Xplore. Restrictions apply.