Professional Documents
Culture Documents
net/publication/331628531
CITATIONS READS
2 1,713
3 authors:
H. El-taweel
Helwan University
2 PUBLICATIONS 6 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
Effect of Tyre-to-Sandy Soil Relative Stiffness on Tyre Deformation and Sinkage View project
All content following this page was uploaded by Hassan Metered on 26 May 2019.
(Manuscript Received: 01 November 2018; Revised: 24 November 2018; Accepted: 25 November 2018)
Abstract
In this paper, optimal lumped vehicle passive suspension parameters are estimated using genetic algorithm (GA) to decrease vibration
levels of the vehicle body and passenger seat so that better ride comfort and vehicle stability are achieved. The optimal values of suspen-
sion parameters for minimizing a combined multi-objective function are tuned using GA due to its simplicity, convergence, and robust-
ness. The equations of motion of five-degrees-of-freedom passive half-vehicle suspension system are derived and simulated using
Matlab/Simulink software. Double bumps and random road excitations are used to study the performance of the suspension system in-
cluding bounce and pitch motion. The performance of the optimized passive suspension system using optimal lumped parameters is
compared with the nominal passive to show the efficiency of the proposed optimized suspension system. The simulation results prove
that the optimized passive suspension system using GA can offer significant improvements of ride comfort and vehicle stability.
Keywords: lumped parameters; passive suspension; genetic algorithm; half vehicle model; ride comfort; vehicle stability.
34
H. El-taweel et al./ Journal of Advances in Vehicle Engineering 5(1) (2018) 34-44
knowledge is further exploited in the optimization of vehicle formulation of the equations of motion.
and seat suspension parameters by investigators. In [6], the Parameters , , , , , , , and which
author presented an optimization of a 4-DOF quarter car seat represent the vehicle’s fixed parameters are expressed as front
and suspension system via GA to compute a set of parameters tire mass, rear tire mass, (seat + driver) mass, sprung mass,
to achieve the superlative performance of the driver’s seat. mass moment of inertia of sprung mass, front tires position in
The objective was proposed as the minimization of a multi- relation to the center of mass, rear tires position in relation to
objective function formed by the combination of not only the center of mass, and seat position in relation to the center of
suspension working space and dynamic tire load but also the mass, respectively. While variables , and de-
head acceleration and crest factor (CF). note seat stiffness coefficient, front suspension stiffness coef-
An optimization of a 7-DOF vehicle’s driver with seat sus- ficient and rear suspension stiffness coefficient, respectively.
pension system using GA was presented in [7] to find out the Also, variables , and denote seat damping co-
seat suspension parameters in order to improve the comfort of efficient, front suspension damping coefficient and rear sus-
the driver. The theoretical results from the optimized vehicle’s pension damping coefficient, respectively. In addition, param-
driver with seat suspension model showed significant im- eters and denote front tire stiffness coefficient and
provements in driver’s comfort. GA is also used to optimize rear tire stiffness coefficient, respectively. Further, subscripts
the parameters of a quarter-car suspension model through 1 and 2 represent tire axes. It is also needed to observe that
considering the minimization of the vehicle body vertical ac- seat type is composed of a linear spring and damper. The dy-
celeration in [8, 9]. The results from their models with opti- namic of this model is excited by double bumps and random
mum parameters showed considerable improvements in peak- road excitations. Also, the data employed here for the half
to-peak (PTP) values of vehicle body vertical acceleration vehicle system is listed in Table 1 [10, 11].
against the vehicle with passive assumptions. In fact, applying
optimization techniques in engineering design has been of
great significance particularly in solving complex real-world
design problems. In multi-objective optimization problems,
there are several objectives or cost functions to be optimized
(minimized or maximized) concurrently. These objectives
often conflict with each other so that as one criterion improves,
another deteriorates. Consequently, there is no single optimum
solution that is best with respect to all the objective functions.
Instead, there is a set of optimal solutions, well-known as Pa-
reto optimal solutions [5, 10].
The purpose of this work is to determine the lumped pa-
rameters of the 5-DOF suspension system that minimize a
combined multi-objective function which takes in considera-
tion the peak-to-peak (PTP) and root mean square (rms) val-
ues of the nominated design criteria under different road
irregularity profiles. The rest of this article is organized as
follows: an overview of the half vehicle model is presented
in the section (2) contained brief details of dynamic equations
of motion. Section (3) describes the road excitation profiles.
Optimization procedures using GA is introduced in section (4).
Then the results obtained for different road disturbance inputs
are presented and discussed in section (5). Finally, the conclu-
sions are summarized in section (6). Figure 1.Vehicle vibration 5-DOF model with passive suspension
2. Overview of half vehicle model By applying Newton’s second law to the half vehicle model,
Figure 1 illustrates schematically a five-degrees-of-freedom the equations of motion are:
passive half-vehicle suspension model, which is adopted from
[11]. This suspension model incorporates one sprung mass )1(
that joints to three un-sprung masses (represent tires and driv-
er’s seat). Furthermore, the effect of degrees of freedom, line- )2(
ar motion (in vertical direction for sprung and un-sprung
)3(
masses) and pitching motion for sprung mass, in terms of
acceleration, velocity and movement, are taken into account in
35
H. El-taweel et al./ Journal of Advances in Vehicle Engineering 5(1) (2018) 34-44
( ) ( ̇ ̇ ) )4(
( ) ( ̇ ̇ ) )5(
( ) ( ̇ ̇ ) )6(
̈ )7(
̈ )8(
̈ )9(
̈ ( ) )11(
̈ ( ) )11(
Figure 2(a). Double bumps road excitation
Where, , , and represent vertical seat dis-
placement, vertical displacement of the central gravity of the
sprung mass, vertical displacement of the ends of the sprung
mass and pitching motion, respectively. Moreover, ̇ , ̇
and ̇ denote vertical seat velocity, vertical tires velocity
and vertical velocity of the ends of the sprung mass, respec-
tively. While, ̈ , ̈ , ̈ and ̈ are vertical seat accelera-
tion, vertical acceleration of the central gravity of the sprung
mass, vertical tires acceleration and pitch acceleration, respec-
tively. Finally, and represent the excitations via
uneven road.
applied [14].
Distance 0.4
Distance 1.38 m
Distance 1.36
To generate optimal solutions, GA will apply genetic evolu- probability Pm, the mutation probability, which is fixed before
tion operators such as crossover and mutation for the for- the optimization. Often the mutation probability is very small.
mation of new chromosomes from the existing ones in the Elitism Strategy
population. This process is accomplished by either merging Elite children are the individuals in the present generation
with the best fitness values. These individuals survive to the
the existing ones in the population or by modifying an existing
next generation automatically.
chromosome. The selection method for parent chromosomes In this paper, the initial modeling approach is further opti-
depends upon the fitness value of the parent. This procedure mized using GA to determine the optimal design values of the
guarantees that a better solution will have a higher possibility passive suspension parameters in order to enhance the ride
to procreate and donate their beneficial characteristic to their comfort and vehicle stability. The following parameters are
offspring’s. GA generates not only one potential solution to aimed to be the design variables:
the problem but also a set of potential solutions. This is well- The vehicle main suspension system including stiff-
known as population. ness and damping coefficients.
The simplest form of genetic algorithm involves four types The driver seat suspension system including stiff-
of operators [14]: ness and damping coefficients.
Selection: These design variables should be optimally determined by a
Selection is the process to choose which individual of the combined multi-objective function. The performance criteria,
population will be permitted to survive. A good parent selec- which are of most interest when designing a vehicle suspen-
tion strategy is essential to create good offspring’s. sion system, are ride comfort and vehicle stability. Ride com-
Crossover: fort is a criterion of vertical accelerations of vehicle body and
Crossover generates two new offspring’s through creating passenger seat, should be minimized to reduce the fatigue of
new genetic material by exchanging the genetic material be- passengers and enhance ride comfort. Vehicle stability is a
tween parents. criterion of the relative displacement between tire and vehicle
Mutation: body (suspension working space) and tire deflection, should
The mutation procedure affects random variation upon the be minimized too to enhance vehicle stability [15].
gene of a chromosome. A mutation is performed with the
37
H. El-taweel et al./ Journal of Advances in Vehicle Engineering 5(1) (2018) 34-44
are achieved.
For these reasons, to obtain the optimal design variables In this section, the performance of the optimized passive
values, all these criteria should be minimized. Consequently, suspension system using the optimized lumped parameters
the objective is to minimize a combined multi-objective func- which are determined via GA is compared with the nominal
tion (J), Eqn. (13), formed by the combination of the peak-to- passive suspension. The above-mentioned performance crite-
peak (PTP) values of the performance criteria, to improve the ria are used to quantify the relative performance of these sus-
suspension performance on transient conditions, and the root pension systems. Since the nominal passive suspension is used
mean square (RMS) values of the nominated criteria, to im- as a base-line for comparisons. The optimal vehicle and seat
prove the suspension performance on realistic or random road suspension design parameters for the proposed 5-DOF passive
condition. half-vehicle suspension model that estimated by applying GA
The goal of the objective/cost function is to minimize a is listed in Table 2 and compared with the initial suspension
combined multi-objectives function (J) and it can be de- design parameters.
scribed as follows: The time history of the suspension systems behavior under
double bumps road excitation is shown in Fig. 4. The respons-
( ) ( ) ( ) es of , , , , , and are present-
ed in Figs. (a, b, c, d, e, f, g, and h), respectively. The latter
( ) ( ) ( ) figures show the comparison between the optimized passive
suspension using GA and the nominal passive suspension
( ) ( ) ( ) systems. From these results it is evident proposed optimized
suspension using GA can dissipate the energy due to double
( ) ( ) ( )
bumps excitation very well, cut down the settling time, and
( ) ( ) (13) improve both the ride comfort and vehicle stability. In addi-
tion, Fig. 4 shows that the proposed optimized passive suspen-
sion using GA have the lowest peaks for the nominated design
Where, ( ) , ( ) , ( ) , criteria, demonstrating their effectiveness at enhancing the
( ), ( ), ( ), and ( ) rep- ride comfort and vehicle stability.
resent peak-to-peak values of front suspension working space, Figure 5 represents the responses of the proposed optimize
rear suspension working space, front tire deflection, rear tire passive suspension system using GA under random road exci-
deflection, body acceleration, passenger (seat) acceleration, tation The responses of , , , , , and
and seat travel distance, respectively. Furthermore, are presented in Figs. (a, b, c, d, e, f, g, and h), respec-
( ), ( ), ( ), ( ), tively. According to these figures, just like for the double
( ), ( ) , and ( ) represent root bumps road excitation, the proposed optimized suspension
mean square values of front suspension working space, rear system dissipate the energy due to road excitations very well
suspension working space, front tire deflection, rear tire de- and improve both the ride comfort and vehicle stability.
flection, body acceleration, passenger (seat) acceleration, and The results show that the optimized passive suspension us-
seat travel distance, respectively. ing GA can reduce maximum peak-to-peak of the ,
This above-mentioned a combined multi-objective function , , , , , and by about
is used in this study to find out the optimal values of all vehi- 59.1 %, 64 %, 44 %, 43 %, 35.3 %, 32.7 %, 36 % and 64.2 %,
cle suspension damping coefficients and spring stiffness's for respectively, compared with the nominal passive suspension
the driver’s seat and main suspension system of the 5-DOF system. Figure 6 shows the improvement percentages of PTP
passive suspension model. The optimization procedures using for the optimized passive suspension using GA compared to
GA are summarized in the flowchart as illustrated in Fig. 3. the nominal passive suspension system. The results confirm
that the optimized passive suspension using GA offers a supe-
5. Results and discussion rior performance.
Suspension working space( ), tire deflection ( ), In the case of random road excitation, it is the root mean
body acceleration ( ), pitch angular displacement ( ), square (RMS) values of the , , , ,
seat travel distance ( ), and passenger (driver) accelera- and , rather than their peak-to-peak values, that are rele-
tion ( ) are the major performance criteria in vehicle sus- vant. The results show that, for random road excitation signal
pension design that govern the ride comfort and the vehicle the RMS of the , , , , and were
stability. Ride comfort is directly related to , and improved by about 54.6 %, 60.3 %, 48.6 %, 54.2 %, 29.2 %,
. On the other hand, good vehicle stability requires low 15.4 %, 30.3 % and 61.2 %, respectively. Figure 7 shows the
tire deflection. The structural features of the vehicle also re- improvement percentage of RMS for the optimized passive
strain the amount of within certain limits [12, 15]. The suspension using GA compared to the nominal passive sus-
main target of this study is to minimize the nominated perfor- pension system. The results again confirm that the optimized
mance criteria so that better vehicle stability and ride quality passive using GA can give a superior response in terms of ride
38
H. El-taweel et al./ Journal of Advances in Vehicle Engineering 5(1) (2018) 34-44
190000 294340
(N/m)
190000 294340
(N/m)
31000 55540.32
(N/m)
16812 29183.34
(N/m)
16812 25230.3
(N/m)
830 1529.87
(Ns/m)
1000 2530.25
(Ns/m)
1000 2185.66
(Ns/m)
39
H. El-taweel et al./ Journal of Advances in Vehicle Engineering 5(1) (2018) 34-44
Figure 4(d). Rear Tire Deflection under double bumps road excitation
Figure 4(c). Front Tire Deflection under double bumps road excitation
road excitation
40
H. El-taweel et al./ Journal of Advances in Vehicle Engineering 5(1) (2018) 34-44
Figure 4(g). Passenger Acceleration under double bumps road road excitation
Figure 4(h). Seat Travel Distance under double bumps road excitation
road excitation
41
H. El-taweel et al./ Journal of Advances in Vehicle Engineering 5(1) (2018) 34-44
Figure 5(e). Body Acceleration under random road excitation Figure 5(g). Passenger Acceleration under random road excitation
42
H. El-taweel et al./ Journal of Advances in Vehicle Engineering 5(1) (2018) 34-44
Figure 6. Percentage improvements in PTP values for the optimized passive using GA compared to nominal passive for double bumps road excitation
Figure 7. Percentage improvements in RMS values for the optimized passive using GA compared to nominal passive for random road excitation
43
H. El-taweel et al./ Journal of Advances in Vehicle Engineering 5(1) (2018) 34-44
44