2011 The 6th International Forum on Strategic Technology

Research on Regenerative Braking Control Strategy

of Hybrid Electric Vehicle
Meilan Zhou, Zhaoming Gao, Hao Zhang
College of Electrical & Electronic Engineering
Harbin University of Science and Technology
Harbin, China
strive63@ 163 .com
Abstract-For the shortage of the original strategy of braking
force distribution in hybrid electric vehicle (HE V) simulation
sofware ADVISOR, a new model which the braking share of
every braking force varies on the change of load was built from
the standpoint of dynamics, based on the original regenerative
braking control model. The model is simulated by ADVISOR
sofware in HAFEI sample car and the simulation results show
that the effect of energy recovery in new control strategy is
superior to the original one, emissions of harmful gases are
reduced and the motor efciency is improved signifcantly. The
braking force distribution method is also in accordance with
ECE regulation. So the control strategy proposed in this paper is
proved to be reasonable and effective. This kind of model
expands the scope of the ADVISOR simulation effectively and
facilitates the study of hybrid electric vehicle.
Keywords-hybrid electric vehicle; ADVISOR; regenerative
braking; change of load; control strateg
I. INTRODUCTION
With the rapid growth of vehicle production, automobile
fel consumption accounts for an increasing proportion of all
oil production, so energy utilization efciency is extremely
important in resource shortage of today [I]. Compared with
conventional cars, the important distinction of hybrid electic
vehicles can realize regenerative braking. It not only improves
energy efciency but also greatly saves resources. This paper
established simulation model of regenerative braking force
distribution during the braking process, according to the
ADVISOR sofware and simulated it. The simulation results
show that the control strategy proposed in this paper can
improve the energy recovery effciency and emissions of
harfl gases reduce in the course of driving.
II. THE SIMULATION PROCESS OF ADVISOR
At present, there are two major kinds of car simulation
method: one is forward simulation, another is backward
simulation. The backward simulation can not be used for
dynamic simulation while a car is driving; the forward
simulation is more accurate, but it increases the computational
amount and slows down the simulation speed. However, most
of the simulation sofware adopts one of the simulation
methods, which can not take fll advantage of both simulation
methods. The ADVISOR sofware adopts the hybrid
simulation methods which based on backward simulation and
Supported by Scientifc Research Fund of Heilongiang Provincial
Education Department (o: 11551072).
978-1-4577-0399-71111$26.00 2011IEEE 300
supplemented by forward simulation, so that it can achieve the
simulation more accurate and faster [2-3].
There are three main operations in ADVISOR: input
vehicle parameters, set up simulation parameters and view
simulation results, they are achieved by three GUI interfaces.
A. Input vehicle parameters
The user can set up parameters of different components of
the vehicle via the interface of vehicle parameters. The
perforance curve of engine or transmission varies on different
parameters settings. Of course, the components of the car
should be matched when we set the parameters. Otherwise,
ADVISOR will give the hint that is "Parameters do not match".
B. Sets im ulation parameters
The user can select test scheme and set the simulation
initializing conditions via the simulation parameters setting
interface, such as drive cycle, cycle index and so on. By setting
simulation parameters in this way, it can help to study the
infuence of the parameters on the performance of the motor
and provide a reference for the automotive design.
c Output Simulation results
Via the simulation results interface the user can see the
changes of SOC, gas emissions, changes of speed ratio. The
specifc text descriptions are also can be seen, such as CO, HC
and other gas emissions value and fel efciency. Clicking the
"Energy Use Figure", the user can see the whole process of
energy fow.
III. REGENERATIVE BRING CONTROL STRATEGY IN
ADVISOR
As shown in Fig. 1, it is a default regenerative braking
distribution mode in ADVISOR. From the fgure we can see
that when the speed is higher than 20km/h, the share of
regenerative braking increases with the growth of speed, but
when the speed is lower than 20kmlh, the share of the fiction
braking force increases. This distribution mode makes the
efciency of energy recovery decrease at lower speed [4].
August 22-24,2011
1

0.9
:

0.8
'"

0.7 0
0"
0.6 .
o.
0.5
;
0.4 O
u
'"
0.3
<
b
0.2
. S

0.1 '"
c
0
0 20 40 60 80 100 120
v(kmlh)
Figure 1. The distribution of braking force in ADVISOR
IV. THE NEW BRAKING FORCE DISTRIBUTION STRATEGY
AND ITS MODULES
By the application of the original mode, braking force
distribution mode based on the variation of load is built.
Analysis on the load of car body and car fame has been made
as follows.
As far as the suspended mass system (body and fame), we
built up space right angle coordinate, and the center of mass to
be taken as the origin coordinates. If we regard the system as a
rigid body, it has six degrees of feedom in the suspension
system, and they are displacements along the x, y, z axes and
angular displacements around the three axes. Corespondingly,
there are three forces F" Fy and F z along the x, y, z axes, and
torque T, bending moment M and M which around the three
axes. As shown in Fig. 2.
=
e
.T
Figure 2. The force placed on the vehicle
With the car is in motion, the force acts on the car body as
follows:
F =
a
m
(G
r
+G
e
)
/
=k (G +G
)
=k
"F
II / g m r e m l'
(m = x,y,z)
(\)
In the forula, km is the dynamic load coeffcient in x, y, z
axes, km=an; am is the car acceleration along x, y, z axes; g is
the acceleration of gravity; Gr is itself load of the suspended
vehicle body system, which refers to the components that bear
by elastic element of the suspension. Such as car body, car
301
fame and all the assemblies fxed in the body or the fame and
the gravity of the equipment; Ge is the payload of the car body,
which refers to the gravity of passengers, baggage and cargo in
rated load.
Gr and Ge are static loads. According to the concentrated
load F (unifor load), they distribute properly in i(i =1 ,
2, ... ),which are in the car body and car fame, the equation is as
follows:
G +G ="F
T e 1
(2)
Dynamic load coeffcient mainly depends on three factors:
road conditions, driving conditions (such as speed) and
automobile structure parameters. As these factors are complex,
so the dynamic load coeffcient is diffcult to be deterined by
mathematical analysis. Therefore we take semi-empirical
values combining theoretical studies with experimental
corection as dynamic load coeffcient.
Practice proves that there are two kinds of basic loads
afecting the strength of car body and car fame in the above
loads
A. Symmetric vertical load
As shown in Fig. 3, when the car is driving on the road
section I and II, and the font and rear wheels of a car are
passing the bumps or pits of equal height at the same time,
supposing the vehicle structure is bilateral symmetrical, the
load bearing system is acted on by the force Fzs which is
perendicular to longitudinal axes. It is the main component of
dynamic load under noral driving conditions of vehicle. The
equation is as follows:
(3)
In the forula, kzs is symmetic vertical dynamic load
coefcient.
Figure 3. The analysis of the road on which the car is running
When the font and rear wheels of a car are passing the
bumps of equal height at the same time, the dynamic load
coeffcient kzs is:
h )
A
1+
v2
(4)
In the formula, CI and C2 are respectively synthetic stifess
(N/mm) of font and rear suspension and tires.
C1
= Ctl Cs1 IC Cll + Cs1)
(5)
(6)
In the forulas, Ctl and Cr2 stand for respectively stifess
of the font and rear wheels; Csi and Cs2 stand for respectively
stifess of the font and rear suspension; h is the height of
baricade, usually h adopts lOOmm for noral trucks in the
calculations, for cars and buses h adopts 80mm; A is the
empirical coefcient, it adopts lOOO(k)2; Va is speed(km/h);
Go is the automotive gravity (N).
The calculations and test results show that the symmetrical
vertical dynamic load coefcient can adopt the following
values for all types of car when strength check is perfored:
Car and bus kz
s
= 2.02.5
Truck
Off-road vehicle
B. Asymmetrical vertical load
As shown in Fig.3, there is an altitude difference hn in the
ground point of lef and right wheels, when the car is driving
on road section III and the font and rear wheels are passing
fom the fat road surface to the bump,so the vertical loads
acting on the load bearing system are asymmetry to the
automobile longitudinal axis, as shown in Fig. 4a. At this time
the load can be decomposed into symmetrical vertical force
(bending working situation, as shown in Fig. 4b) and vertical
load Fzn caused by the bump of the road, and it is symmetry for
the automotive longitudinal axes. And the torque Tx of car body
around the x axes is fored by F/J and FIR (torsion working
situation, Fig. 4c), the equations are as follows:
(7)
(8)
In the forulas, FJL and FIR stand for the forces acting on
the lef and right font wheels respectively; BI is the font gauge,
B2 is rear gauge; Kzn is asymmetric vertical dynamic coefcient;
Kzn=1.3(car and bus), Kzn=1.3 (truck), Kzn=1.8(off-road vehicle)
[5].
Figure 4. Asymmetrical vertical load
302
It should be pointed out that the exteral load which the
vehicle suffered in motion has random nature, such as fatigue
load (random load). It can be negligible because of its less
infuence on the aspects of the study. When the car is driving
on the road, most of the loads acting on the car are
asymmetrical vertical loads. The simulation results show that
there is no effect on the result to change the gain constant value
in the module.
V SIMULATION RSULTS AND ANALYSIS
In order to compare the improved control strategy with the
original control strategy in ADVISOR, and facilitate the
comparison with the actual situation. The Hafei HF3A sample
car as a prototype is chosen to research, the specifc parameters
are shown in Table I . Simulation road condition is
CYC_UDDS, as shown in Fig. 5 [6].
TABLE!. MAIN SIMULATION PARAMETERS
Whole-car parameters
Height of
Distance
Wole-car Wole-car from
qualit qualit
gravit
Weelbase gravit
center
(ully load) (no load)
(ully load)
(mm) center to
(kg) (kg
(mm)
front-ale
(mm)
1580 1280 508 2600 1470
Distance
from
Wnd Wndard Coefcient Weel
gravit
center to
resistance acreage of roling radius
front-ale
coefcient (n) resistance (mm)
(mm)
1130 0.335 2.31 0.009 289
CYC UDDS
Figure 5. CYC_UDDS condition chart
According to the above analysis, the original module in
ADVISOR is modifed and the new regenerative braking
control module is established. The modifed module of braking
force in backward path is shown in Fig. 6 and the modifed
module of braking force in forward path is shown in Fig. 7.
Constant
Constantt
Figure 6. The modifed regenerative braking module in the backward path
Figure 7. The modifed regenerative braing module in the forward path
The simulation results show that the strategy adopted by
this paper has obvious superiority over the original strategy in
the aspect of braking energy recovery. In the whole cycle, the
battery energy storage is increased, the exhaust emission is
reduced and the motor efciency is also signifcantly improved.
As shown in Fig. 8 and Table II .
07
The simulation result of the secondary development
t =
.,
g 0.65
i
-
The simulation results of the original model
200 400 600 800
Time(s)
1000
Figure 8. The change of SOC
1200 140[1
TABLE II. THE COMPARISION OF SIMULATION RESULTS
The original ADVISOR The improved
strateg strateg
The average power
generation efciency of
57.5828 70.1458
motor
(%)
The average electric
37.5805 39.3337
efciency of motor
(%)
Energy recover
1377 1586
(KJ)
HC 0.52083 0.01703
Exhaust
Emission
CO 2.3608 2.3374
(grams/mile)
NO, 0.40698 0.39844
303
It can be seen fom the Fig. 9 that the curve of braking
force distribution of HF3A is above the synchronous
attachment coeffcient, the braking force distribute in a closely
way to the ideal curve I. And it is higher than the minimum
distribution curve, which meet the ECE regulation. Therefore,
this distribution method is consistent with the distribution
requirements of braking force regulation [7].
T eurc(rull load)
B eure
T eure(no load)
ECE re ulation
] II 12 13
Braking rorcc or the rront whecl Fu 1 (kN)
Figure 9. Braking force distribution curve
VI. CONCLUSION
Regenerative braking is a signifcant distinction between
hybrid electric vehicles and conventional vehicles. In the study
of hybrid electric vehicles, regenerative braking has become an
important technique in reducing energy consumption and
improving the driving range. This paper establishes a new
model which the braking share of every braking force varies on
the change of load fom the standpoint of dynamics, based on
the original module of regenerative braking in ADVISOR
sofware. Simulation results show that the model built in this
paper is correct, and the simulation fnction of ADVISOR is
extended, which facilitates the study of the regenerative
braking of hybrid electric vehicle.
RFERENCES
[1] Fangfang Lei. Development of New Energy Vehicles in China [1].The
Motor Engineer, 2009(5) : 12-14
[2] Xiang Zhang, Han Zhao, Lijun Qian, etc. Combined Backward/ Forward
Simulation Approach in the ADVISOR [J].Computer Simulation, 2005,
22(2):203-206.
[3] Guozhu zhao, Minxiang Wei, Zhenglin Yang. Development of a Module
for the Regenerative Braking in ADVISOR Based on the Braking
Stability [.]. Mechanical and Electronic Press, 2007(6): 10 -14.
[4] Xiao Ling , Hodgson ..W . Modeling and Simulation for Hybrid
Electric Vehicles[J] . IEEE Transactions on Intelligent Transportation
Systems, 2002: 244-251.
[5] .inling Huang. Car Body Design [M]. Beijing: Machinery Industry Press,
2007.
[6] Xiaohua Zeng, Qinghua Wang, etc. The Development of HEY Control
Strategy Module Based on ADVISOR2002 Sofware [1]. Automotive
Engineering, 2004 26(4): 394-396
[7] Liang Chu, Mingli Shang, Yong Fang . Braking force distribution
strategy for HEY based on braking strength