A Mathematical Model For Coupled Vibration System of Road

JID: APM
ARTICLE IN PRESS [m3Gsc;August 14, 2015;11:35]
Applied Mathematical Modelling 000 (2015) 119
Contents lists available at ScienceDirect
Applied Mathematical Modelling

journal homepage: www.elsevier.com/locate/apm
A mathematical model for coupled vibration system of road

vehicle and coupling effect analysis
Junhong Zhang, Peng Guo, Jiewei Lin, Kainan Wang
State Key Laboratory of Engines, Tianjin University, 92 Weijin Road, Tianjin 300072, China
a r t i c l e i n f o a b s t r a c t
Article history: A mathematical model describing the dynamics of road transport vehicle is proposed in this
Received 26 May 2014 paper. The system consists of a cab, a carriage, a chassis, the road, and mounts/suspensions
Revised 14 April 2015
between them. The coupled correlations between substructures are studied using the state
Accepted 8 July 2015
space theory. A factor, coupling coecient, for describing the coupling effect in the subsys-
Available online xxx
tem and a factor, vibration attenuation coecient, for characterizing the vibration reduction
Keywords: of the substructures are derived. The joint location, the stiffness, and the damping of the
Vehicle mount/suspension, as well as the mass and inertia moment of the substructures are consid-
Mathematical modeling ered as design variables affecting system vibration coupling and reduction. The contribution
Coupled system of design variables is analyzed using Latin hypercube sampling and quadratic regression, top
Dynamics ten contributing variables are obtained, and their effects on the subsystem coupling are an-
Sensitivity analysis alyzed and discussed. The back propagation neural network is employed to investigate the
nonlinear correlation between subsystem coupling and substructure vibration reduction. The
results show the cab vibration reduction achieves optimum when the cabchassis coupling is
high and the carriagechassis coupling is low. However, the optimal carriage vibration reduc-
tion requires the exact opposite coupling correlations.
2015 Elsevier Inc. All rights reserved.
1. Introduction
Vibration of cab and carriage has great inuence on ride comfort and goods safety, as well as vehicle reliability. The vibration
on cab oor, seat, backrest, and steering wheel has been reduced as low as possible to enhance drivers comfort. On the other
hand, however, the vibration of carriage is usually considered as an unimportant factor in the ride comfort. For some specialized
purpose, such as medical purpose delivery or special equipment delivery, the vibration of carriage requires a much higher level
of concern.
Transport vehicle is a very complex system consisting of cab, carriage, chassis, wheel, and kinds of joints between them
such as mounts, suspensions, and tires. The excitation of road roughness acts on the tires and is transmitted to the wheel and
suspension, then the chassis or frame, nally the cab and carriage through the mounts. Taking the sprung mass of vehicle as a
multi-supported system, the coupling effects between the substructures strongly affect the vibration transmission and vehicle
response. An inappropriate coupling intensity might lead to a signicant magnication of vibration response of cab or carriage
to some operating conditions (road pavement, frequency range or speed). However, it is dicult to investigate or design the
coupling effect in those subsystems. The relation between vibration reduction and coupling intensity is also dicult to establish
for complex interactions and large number of factors.

Corresponding author. Tel.: +8613752356026; fax: +862227407851.
E-mail address: linjiewei@tju.edu.cn (J. Lin).
http://dx.doi.org/10.1016/j.apm.2015.07.012
S0307-904X(15)00435-7/ 2015 Elsevier Inc. All rights reserved.
Please cite this article as: J. Zhang et al., A mathematical model for coupled vibration system of road vehicle and coupling effect
analysis, Applied Mathematical Modelling (2015), http://dx.doi.org/10.1016/j.apm.2015.07.012
JID: APM
2 J. Zhang et al. / Applied Mathematical Modelling 000 (2015) 119
Nomenclature
Symbol
A interaction matrix of the substructure state
B input matrix of system
C damping matrix
D state matrix
E displacement transform matrix
F force matrix
G transfer function
H inspection matrix
I identity matrix
I moment of inertia
K stiffness matrix
L distance from joint to reference point
M mass matrix
q global coordinate
Q state variable
R position vector
T coordinate transform matrix
x, y, and z translational displacement in x, y, and z directions
, , and rotational displacement in x, y, and z directions
, , and precession angle, nutation angle, and rotation angle
coupling coecient of subsystem
vibration attenuation coecient of substructure
Superscript and subscript

B basis
S sprung mass
M mount
sus suspension
cab cab
car carriage
cha chassis
f front
r rear
The fundamental purpose of a vehicle suspension/mount system is to act as a vibration isolation among the road, wheel,
chassis, frame, driver and passenger, and cargo. Many researchers have done a lot of effective work on the suspension and mount
design, improvement and optimization. In vehicle suspensions, most investigations focus on multi-body dynamic modeling and
optimization design of semi-active and active dampers and control model [17]. In powertrain/engine mounts, optimizations
of passive mount in shape and arrangement, and design and development of active/semi-active dampers are the two main in-
terests. Refs. [813] investigated optimal designs of passive engine mounts by parameterizing mount factors which inuence
the shock/vibration isolation and reduction, and the optimization procedures were carried out using analytical method, nite
element method, or genetic algorithm. In order to control the vibration transmissibility from engine to chassis and cabin, great
designs or optimizations have been accomplished on semi-active engine mount [14,15] and active mount systems [1618]. In
addition to the isolation of vibration sources (road roughness and engine vibration), to reduce the vibration transmitted into cab
and carriage of road transport vehicle, mount system of the cab and carriage have been studied as well [1929].
Although many valuable investigations on vehicle suspension, cab mount, engine mount, and body mount have been pre-
sented, the objective of the optimal designs usually focused on only one substructure, such as cab ride comfort or noise, vi-
bration transmitted from the powertrain or road excitations, and body vibration. However, a vehicle includes many subsystems
and substructures, and the correlations between them are so complicated that an adjustment of one subsystem or substructure
may introduce several uncertain or unexpected inuences on the others. For example, an improvement of cab mount that can
reduce cab vibration effectively, however, may increase the acceleration of carriage at the same time, which is not considered in
most studies. As a result, to identify the coupling correlations between the subsystems clearly would be a precondition of vehi-
cle component optimization. It is important to nd out the effect of design parameters on the coupling relationships between
components, substructures, and subsystems.
In this paper, a dynamic model of vehicle, which is taken as a coupled vibration system consisting of the cab, the carriage,
the chassis, mounts, suspensions, wheels and tyres, is developed. The equations of motion are derived and described in the
state space form in order to nd out functions representing the coupling effect and the vibration reduction of the subsystem.
JID: APM
J. Zhang et al. / Applied Mathematical Modelling 000 (2015) 119 3
Fig. 1. Schematic diagram of a multi-supported system.
Two factors, coupling coecient and vibration attenuation coecient, are proposed. Inuences of various design parameters
on subsystem coupling and vibration reduction are analyzed. With sensitivity analysis, top ten design parameters that affect
the coupled vibration most are identied. The effect of design parameter on coupling coecient and the nonlinear relationship
between subsystem coupling and substructure vibration attenuation are explored and discussed in detail.
2. Mathematical model
2.1. Equation of motion
The schematic diagram of a multiple supported system, as shown in Fig. 1, is constructed with one basis/unsprung mass
(supporting structure) and multiple sprung masses (supported structures).
The global coordinates of the basis, qB , and the sprung mass, qS , are dened as
qB = (xB , yB , zB , B , B , B )
T
(1)
qS = (xS , yS , zS , S , S , S ) ,
T
(2)
where x, y, and z are translational coordinates, , , and are rotational coordinates, and subscripts B and S are for the basis
and sprung mass. The position vectors of the mount and sprung mass, RM and RS , with respect to the basis are given as
R M = ( xM , yM , z M ) T (3)
R S = ( xS , yS , z S ) T , (4)
where R is the position vector, and the subscript M is for the mount. The displacement of mount on the basis side can be written
as

xB

yB

1 0 0 0 zM yM

zB
RBM = 0 1 0 zM 0 xM = E B qB , (5)
xM B

0 0 1 yM 0
B

B
where EB is the displacement transform matrix of the basis. The displacement of mount on the sprung mass side is written as

xS

yS

1 0 0 0 zM zS (yM yS )

zS
RM = 0 1 0 (zM zS )
S
0 x M xS = E S qS , (6)
0 0 1 yM y S (xM xS ) 0 S

S

S
where ES is the displacement transform matrix of the sprung mass. Then, the deformation of mount is given by
RM = RSM RBM = E S qS E B qB . (7)
The principle stiffness and damping of mount are dened as

KM,x 0 0
KM = 0 KM,y 0 (8)
0 0 KM,z
JID: APM

CM,x 0 0
CM = 0 CM,y 0 , (9)
0 0 CM,z
where KM, x , KM, y , and KM, z are stiffness components in the x, y, and z directions, CM ,x , CM ,y , and CM ,z are damping components in
the x, y, and z directions. Then the principle force can be dened as
F M = K M T R M , (10)
where T is the transform matrix from the local coordinates to the global coordinates, and is given by
T
cos cos sin cos sin cos sin sin cos cos sin sin
T = sin cos cos cos sin sin sin + cos cos cos cos sin , (11)
sin sin sin cos cos
where , , and are the precession angle, nutation angle, and rotation angle, respectively.
By transforming the force to the local coordinates, the spring force and the damping force of mount can be obtained as
F kM = E M T T T K M T (E S qS E B qB ) (12)
F cM = E M T T TC M T (E S qS E B qB ), (13)
where F denotes the force matrix, the superscript k is for the stiffness and c is for the damping. The equations of motion of
the basis and the sprung mass are derived as

M B qB + E B T T TC M T E B qB + E B T T T K M T E B qB , = F B + E B T T TC M T E S qS + E B T T T K M T E S qS
(14)

M S qS + E S T T TC M T E S qS + E S T T T K M T E S qS , = F S + E S T T TC M T E B qB + E S T T T K M T E B qB
(15)
where FB and FS are the external forces acting on the basis and the sprung mass, respectively.
2.2. State space description
Based on the state space theory [30], the state variable, Q, is given as

q
Q= . (16)
q
According to the equations of motion of the vibration system, the excitations of substructure consist of external load on the
system and internal force transmitted by the elastic mount. The system input of the state space function can be divided into an
external part and a coupled part with respect to the adjacent substructure. The system input can be written as

Q B = DB Q B + ABS Q S + BBU B
, (17)
Q S = DS Q S + ASB Q B + BSU S
where DB and DS are the state matrices of basis and sprung mass, ABS and ASB are the interaction matrices of substructure states,
BB and BS are the input matrices of basis and sprung mass, and UB and Us are the generalized internal forces of basis and sprung
mass. DB , Ds , ABS , and ASB take the forms as
T T
M 1
E B T T TC M T E B M 1 EB T K MT EB
DB = B B , (18)
I 0
1 T T T T
M S E S T C M T E S M 1 ES T K MT ES
DS = S , (19)
I 0
1 T T T T
B M B E B T C M T E S M 1 EB T KMT ES
AS = B , (20)
0 0
1 T T T T
S M S E S T C M T E B M 1 ES T K MT EB
AB = S . (21)
0 0
JID: APM
Then, the state equation of the coupled vibration system is given by

DB ABS1 ABSi

Q B

Q BB
B
U

Q S1 AS1 D
B S1
B
Q S1

BS1

U S1
= . + . , (22)
..
.. . .. .. ..

.
. .

. .

Q Si Q Si BSi U Si
ASi
B DSi
in which the state matrix of coupled vibration system, Dcoupled , can be split into two components as
n
Dcoupled = Duncoupled + Ai , (23)
i=1
where Duncoupled is the state matrix of the uncoupled substructures, and Ai is the interaction matrix of the coupling basis and
sprung mass. Duncoupled and Ai are given by

DB
DS1
Duncoupled = , (24)
..
.
DSi

0 ASi,B 0
..
.

Ai =
AB,Si
..
.
.
(25)
.
.
.
0 0
2.3. Dynamic characteristic
In order to analyze the inuence of substructure variable on the coupling system characteristic, a coupling coecient, i ,
which is the ratio of Frobenius norms of Ai and Dcoupled , is introduced and taken the form of
A
i = i F . (26)
Dcouple
F
The coupling coecient quanties the correlation of different substructures of a subsystem. A higher value of the coupling
coecient represents a stronger correlation between different substructures. Furthermore, an inspection matrix, H, is introduced
to evaluate the vibration attenuation of the subsystem, which is dened as

H (D) = exp DT t exp (Dt )dt. (27)
0
Using the ratio of two-norm of H, the vibration attenuation coecient is dened as

H (Duncoupled + nj=1 A j )
i = .
2
H (Duncoupled + nj=1, j=i A j )
(28)
2
Because the inspection matrix, H, can be solved using the Lyapunov equation [31],
DT H (D)+H (D)D + I = 0, (29)
so that the expression of i is given by

T

0 exp D + nj=1 A j t exp D + nj=1 A j tdt

i = T
2
, (30)
n n
0 exp D + A
j=1, j=i j t exp D + A
j=1, j=i j tdt
2
where i is employed to quantify the relative vibration reduction of different substructures. If the vibration attenuation coef-
cient of a substructure is high, it means the performance of mounts is good and the vibration response would be relatively
low. However, for a multiple supported system, it is dicult to achieve peak values of vibration attenuation coecient of every
substructure at the same time. As a result, the object of the optimal design of the coupled system may be a compromise between
different vibration reduction effects of the substructures.
JID: APM
Fig. 2. Schematic diagram of the vehicle model.
2.4. Vehicle model
The vehicle model in this paper consists of cab, carriage, chassis, axles, mounts, and suspensions. As shown in Fig. 2, the
presented model represents a three-axle road transport vehicle. There are four spring-dampers for cab mounts, eight for carriage
mounts, four for suspensions, and six springs for tyres. The degrees of freedom (DOFs) in the vertical, rolling, and pitching
directions are considered. The road excitation is unsymmetrical for the left and right wheels. In the vehicle vibration analysis,
the chassis is supported by road, and the cab and carriage are thought as the sprung mass.The state variables of the substructures,
including chassis, cab and carriage, are written as
T
Q B = zcha , cha , cha , zcha , cha , cha , (31)
T
Q S1 = zcab , cab , cab , zcab , cab , cab , (32)
T
Q S2 = zcar , car , car , zcar , car , car . (33)
(1) Chassis
The chassis is supported by four suspensions and supports the cab and carriage with four and eight mounts, respectively.
According to the state matrix, Eq. (18), the dynamic characteristic of chassis is given by

M cha 1C cha M cha 1 K cha
Dcha = , (34)
I 33 033
where Mcha , Ccha , and Kcha are the mass, damping, and stiffness matrices of chassis, respectively. Those three matrices are written
as

mcha 0 0
M cha = 0 Ix,cha 0 , (35)
0 0 Iy,cha
C cha = E susT sus cabT cab carT car

cha C sus E cha + E cha C cab E cha + E cha C car E cha , (36)
K cha = E susT sus cabT cab carT car

cha K sus E cha + E cha K cab E cha + E cha K car E cha , (37)
JID: APM
Table 1
Design variable and range of cab-carriage-chassis coupling system.
Variable Denition Unit Range
Position
l1 Longitudinal distance from front suspension joint to chassis center of gravity (CoG) m 2.883.52
l2 Longitudinal distance from middle and rear suspension joint to chassis CoG m 0.78750.9625
l3,4 Lateral distance from suspension joint to chassis CoG m 0.810.99
l5 Longitudinal distance from cab CoG to chassis CoG m 2.43362.9744
l6 Longitudinal distance from carriage CoG to chassis CoG m 0.180.22
l11,12 Longitudinal distance from mount of cab to cab CoG m 0.87931.0747
l13,14 Lateral distance from mount of cab to cab CoG m 0.720.88
l21,22 Longitudinal distance from middle mount of carriage to carriage CoG m 0.991.21
l23,24 Longitudinal distance from mount of carriage to carriage CoG m 1.82.2
l2528 Lateral distance from mount of carriage to carriage CoG m 0.91.1
Inertia
mcab Mass of cab kg 20522508
Ix ,cab Moment of inertia of cab about x-direction kgm2 282.6345.4
Iy ,cab Moment of inertia of cab about y-direction kgm2 430.2525.8
mcar Mass of carriage kg 900011000
Ix ,car Moment of inertia of carriage about x-direction kgm2 1131.31382.7
Iy ,car Moment of inertia of carriage about y-direction kgm2 1367.11670.9
Damping
Ccab1,2 Damping of front mount of cab Ns/m 47705830
Ccab3,4 Damping of rear mount of cab Ns/m 54906710
Ccar18 Damping of mount of carriage Ns/m 11701430
Stiffness
Kcab1,2 Stiffness of front mount of cab N/m 4320052800
Kcab3,4 Stiffness of rear mount of cab N/m 2700033000
Kcar18 Stiffness of the mount of carriage N/m 43470005313000
Ksus1,2 Stiffness of front suspension N/m 535851654929
Ksus3,4 Stiffness of middle and rear suspension N/m 13969801707420
where C is the damping, K is stiffness, and the subscript sus is for suspension, cab is for cab, car is for carriage. The E is the
transform matrix between different substructures and are given by

1 l3 l1
1 l3 l1
E sus
cha =
(38)
1 l4 l2
1 l4 l2

1 l13 (l11 + l15 )
1 l13 (l11 + l15 )
E cab
cha =
, (39)
1 l14 (l15 l12 )
1 l14 (l15 l12 )

1 l18 (l16 l20 )
1 l18 (l16 l20 )

1 l23 (l21 l20 )

(l21 l20 )
1 l23
E cha =
car
, (40)
1 l24 l22 + l20

1 l24 l22 + l20

1 l19 l17 + l20
1 l19 l17 + l20
where l is the design parameter which is listed in Table 1.

(2) Cab
The cab is supported by the xed basis on four mounts. According to the state matrix, Eq. (19), the dynamic characteristic of
the cab is given by

M cab 1C cab M cab 1 K cab
Dcab = , (41)
I 33 033
JID: APM
where Mcab , Ccab , and Kcab are the mass, damping, and stiffness matrices of the cab, respectively. The matrixes are written as

mcab 0 0
M cab = 0 Ix,cab 0 , (42)
0 0 Iy,cab
C cab = E chaT cha
cab C cab E cab (43)
K cab = E chaT cha

cab K cab E cab , (44)

1 l13 l11
1 l13 l11
E cha =
l
cab . (45)
1 l 14 12
1 l14 l12
(3) Carriage
The carriage is supported by the xed basis on eight mounts. According to the state matrix, Eq. (19), the dynamic characteristic
of the carriage is expressed as

M car 1C car M car 1 K car
Dcar = , (46)
I 33 033
where Mcar , Ccar , and Kcar are the mass, damping, and stiffness matrices of the carriage, respectively. The matrixes are written as

mcar 0 0
M cab = 0 Ix,car 0 , (47)
0 0 Iy,car
C car = E chaT cha
car C car E car (48)
K car = E chaT cha

car K car E car , (49)
1 l18 l16

1 l 18 l16
1 l23 l21

1 l23 l21
E cha = . (50)
car
1 l24 l22
1 l24 l22

1 l19 l17
1 l19 l17
(4) Chassiscabcarriage coupling system
As mentioned above, the coupling effect of subsystem cab is described using interaction matrix. The expressions of the effect
of chassis on the cab, Acab cha
cha , and the effect of cab on the chassis, Acab , are given by

M cab 1C cab M cab 1 K cab
Acab
cha =
cha cha (51)
033 033

Acha
cab =
cab cab , (52)
033 033
where C cab cha cab cha

cha , C cab , K cha , and K cab are the damping and stiffness coupling matrices.
In the case of chassiscarriage coupling, the interaction matrixes are obtained as

M car 1C car M car 1 K car
Acar
cha = cha cha (53)
033 033

Acha
car =
car car , (54)
033 033
where the damping and stiffness matrixes are dened in the same way as the chassiscab coupling case.
The input matrix of substructure, B, represents the effect of external or internal excitation on the system state. The input
matrices of each substructure are expressed as

M cha 1
Bcha = , (55)
033
JID: APM

M cab 1
Bcab = , (56)
033

M car 1
Bcar = , (57)
033
The uncoupling system matrix can be written as

Dcha 066 066
Duncoupled = 066 Dcab 066 . (58)
066 066 Dcar
The coupling characteristic matrixes can be obtained as

066 Acha
cab 066
Acab = Acab
cha 066 066 (59)
066 066 066

066 066 Acha
car
Acar = 066 066 066 . (60)
Acar
cha 066 066
Combining the Eqs. (55), (56), and (57), the system input matrix can be written as

Bcha 063 063
B= 063 Bcab 063 , (61)
063 063 Bcar
Combining the Eqs. (58), (59), and (60), the coupling system matrix can be given by

Dcha Acha
cab Acha
car
Dcoupled = Acab
cha Dcab 066 . (62)
Acar
cha 066 Dcar
(5) Transfer function

By taking the state variable of the cabcarriagechassis coupling system as the output, one can obtain
1
Q ( ) = I Duncoupled AS BU ( ), (63)
where is the Laplace operator [32]. Extending Eq. (63) gives

Q cha Gchacha Gcabcha Gcarcha U cha
Q cab = Gchacab Gcabcab Gcarcab U cab , (64)
Q car Gchacar Gcabcar Gcarcar U car
where G is the transfer function between the two substructures. For the cabcarriagechassis coupling system, the road excita-
tion transmitted by suspension is the main external force. The transfer functions of each substructure can be written as
!1
Gchacha = ( I Dcha ) Acha
cab L(exp Dcab t )Acha Acar L(exp Dcar t )Acha
cab cha car
Bcha , (65)
" 1 1
#1
Gchacab = ( I Dcha )Acab
cha ( I Dcab ) Acha
cab Acar L(exp Dcar t )Acha Acha
cha car cab
( I Dcab ) Bcha , (66)
1 cab car 1
!1
Gchacar = ( I Dcha )Acar
cha ( I Dcar ) Acha
cab L(exp Dcab t )Acha Acha ( I Dcar ) Acha
car Bcha , (67)
where L() is Laplace transformation [32].
3. Design variables
Loading positions on the chassis are shown in Fig. 3. The design variables of the vehicle are listed in Table 1.
JID: APM
Fig. 3. Schematic diagram of loading position of force on chassis.
Fig. 4. Comparison of calculated and measured vibration responses at carriage oor.
4. Analysis and result
4.1. Model validation
In order to validate the analytical model of vehicle, a road test of a three-axle truck was performed according to ISO 26311.
The test was carried out in a test eld on a B-class asphalt pavement, the carriage oor center acceleration was measured at ve
speeds, 40 km/h, 50 km/h, 60 km/h, 70 km/h, and 80 km/h.
In the simulation, the power spectrum density (PSD) and the spatial domain function of road roughness is generated using the
harmonic superposition method according to ISO 8608. The equations of motion of the truck carriage vibration are solved using
the Newmark- method, and the vertical accelerations of the carriage are calculated under the ve speeds mentioned above.
JID: APM
Table 2
Contribution of design variable to the coupling coecients of cab and carriage.
Order Cab Carriage
Design variable Contribution % Design variable Contribution %
1 l11 15.49 l4 15.60

2 Kcab1(2) 12.20 l17 12.84
3 Iv ,cab 6.59 Iu ,cha 12.16
4 I v ,car 5.06 I v ,car 10.00
5 Iu ,cha 4.36 Kcar7(8) 9.44
6 l17 4.15 Ksus,r1(2) 8.06
7 l16 4.00 l3 5.93
8 Kcar7(8) 3.68 l22 4.44
9 Kcar1(2) 3.57 Kcar5(6) 3.97
10 l24 3.13 Ksus,f1(2) 3.28
Fig. 5. Effects of position variable of cab mount on coupling coecients.
The comparison between tested and calculated results are shown in Fig. 4. As we can see, the analytical results are well agreed
with the vibration response measurement.
4.2. Latin hypercube sampling
According to the transfer functions, shown in Eqs. (65), (66) and (67), the response of each single substructure in the cab
carriagechassis coupling system to the road excitation is coupled with other substructures. Thus, parameter optimization of
one substructure affects the coupling effects of other substructures. In order to quantify the main effects of the design parameter
on both global responses and coupling effects, a parameter analysis is carried out. The design parameters for cabcarriage
chassis coupling system contain mass, inertia moment, stiffness, damping, and joint location. Due to the huge amount of design
parameter, Latin hypercube sampling [33] is used for minimizing required recourses while maximizing the information gained,
as explained and fully detailed in Ref. [34].
According to the engineering experience, the design parameter range is set to be 10 percentage of the baseline, and the
sampling scale is 10000. In this paper, all design variables are normalized in order to estimate the contributions fairly. The
quadratic regression method is employed to analyze the contribution.
4.3. Back propagation neural networks
The objective of applying the back propagation neural network (BPNN) technique [35] is to nd out the relationship between
subsystem coupling coecients, cab and car , and vibration attenuation coecients, cab and car .
In this paper, the BPNN consists of three layers: input, hidden, and output layers. There are two input neurons which are the
coupling coecient of cab and the coupling coecient of carriage. The vibration attenuation coecient of cab and the vibration
attenuation coecient of carriage are used as the two outputs of the network. Considering the computation accuracy and time
cost comprehensively, the neuron number in the hidden layer is set to be 10. The normalized data set consists of 10000 groups
JID: APM
(a)
(b)
Fig. 6. Effects of position variable of carriage mount on (a) coupling coecient of cab, and (b) coupling coecient of carriage.
of sample obtained from the Latin hypercube sampling. The training set consists of 9000 groups and the testing set 1000 groups.
LevenbergMarquardt algorithm [36] is employed to train the network parameters with the train error limit of 108 , and the
mean squared error is used to estimate the network performance.
4.4. Contribution analysis
According to the contribution analysis results, top 10 design variables contribute to the two sprung substructures coupling
coecients are listed in Table 2.
From the contribution of the design parameters, it can be seen that the mount position, stiffness, and inertial parameter
have signicant inuences on the coupling effects of the substructures. The damping parameters do not appear in the top 10
JID: APM
Fig. 7. Effects of position variable of suspension on coupling coecients.
Fig. 8. Effects of position variable of substructure centroid on coupling coecients.
variables for both two sprung substructures. Then, the relationship between the coupling coecient and the vibration attenua-
tion coecient of the substructure are studied.
4.5. Effect of suspension/mount position on coupling coecient
In this discussion, all calculations are performed with the same design parameter change (10% from the baseline value) in
order to nd out their contribution to the coupling effect on the subsystem (cabchassis and carriagechassis).
The relationship between subsystem coupling coecient and cab mount position is shown in Fig. 5. The coupling coecient
of the carriagechassis is much larger than the cabchassis indicating a stronger coupling effect of the carriagechassis. It also
shows that the coupling vibration between the carriage and the chassis is more dicult to control. It can be found that the
coupling coecient of the cab, cab , raises 30.8% with increased longitudinal distance between the front mount of and the
center of gravity (CoG) of the cab, l11 , which implies the coupled vibration between the cab and the pitching of chassis could
be improved by increasing this distance. Due to the suppression on the pitching vibration of chassis, increasing the longitudinal
distance between the rear mount of cab and the cab CoG, l12 , will decrease the coupling coecient, cab , by 4.3%. The correlations
between the lateral distances, l13 and l14 , and the coupling coecient of cab, cab , are both positive. When the lateral distances,
l13 and l14 , are increased by 20%, the coupling effect of cab will increase by 5.2%. The inuence of these four position parameters
JID: APM
Fig. 9. Effects of stiffness variable of suspension on coupling coecients.
Fig. 10. Effects of stiffness variable of cab mount on coupling coecients.
of cab mount on the coupling coecient of carriage, car , is fully reversed, and the effects are all less than 1% with the same
parameter changes (20%) in the four position parameters. It should be noticed that, for both two coecients, the sensitivity of
longitudinal distance is lower than the lateral one.
The relationship between the subsystem coupling coecient and carriage mount position is shown in Fig. 6. The coupling
effect of carriagechassis is much more than that of the cabchassis. The analytical result shows a negative correlation between
the coupling coecient of cab, cab , and distances between mounts and carriage CoG, while the tendencies for car are opposite.
The sensitivities of longitudinal distances from front and rear mounts to carriage CoG, l23 and l24 , are 7.3% and 7.5%. They are
a little higher than the distances from middle mounts, l21 and l22 , which are both 5.63%. Notably, the sensitivities of middle
mounts of carriage on carriagechassis coupling are symmetrical (5.63% for lateral and 5.59% for longitudinal). Furthermore, in
cabchassis condition, those four middle mount positions almost have the same effect, except the sensitivity of l22 is a little
smaller. The sensitivity of the relative position of carriage mount with respect to carriage CoG on the cabchassis coupling effect
is signicantly stronger than that on the carriagechassis coupling.
The relationships between subsystem coupling coecients and joint position of suspension and chassis are shown in Fig. 7.
Similarly, the coupling effect of carriage and chassis is higher than the cab and the chassis. It can be seen that both two coupling
coecients, cab and car , decrease with an increasing values of suspension span, l3 and l4 . While the lengths of wheelbase, l1
and l2 , have little inuence on the two coupling coecients. The reason is that the lateral span of suspension mainly affects the
JID: APM
Fig. 11. Effects of stiffness variable of carriage mount on coupling coecients.
Fig. 12. Effects of inertia variable of cab on coupling coecients.
rolling vibration of chassis, which increases with increasing distance, so that the coupling effect between chassis and road will
enhance, and the cabchassis coupling and the carriagechassis coupling will reduce at the same time. On the other hand, the
sensitivities of span and wheelbase on the two coupling coecients are small (both cab and car change less than 1% with a 20%
variation of l1 , l2 , l3 and l4 ). As a result, the suspension span and wheelbase are able to affect the coupling effect of subsystem but
are only suitable for a minor adjustment.
The relationships between subsystem coupling coecients and substructure CoGs are shown in Fig. 8. The coupling effect
of carriagechassis is much stronger than the cabchassis subsystem. Among the inuences on subsystem coupling, the effect
of the distance from cab CoG to chassis CoG, l5 , on the coupling of the cabchassis is the most sensitive one (3.2%). The result
shows that the coupling coecient, cab , enhances when the cab moves away from the chassis CoG, which means an extension
of vehicle total length will introduce more diculties in decoupling as well as the vibration control. Apart from that, the rest
effects of the two longitudinal positions of cab and carriage on the subsystem coupling are very insensitive.
JID: APM
Fig. 13. Effects of inertia variable of carriage on coupling coecients.
Fig. 14. Effects of inertia variable of chassis on coupling coecients.
4.6. Effect of suspension/mount stiffness on coupling coecient
The relationship between coupling coecients and stiffness of suspension, as shown in Fig. 9, denotes that both coupling
effects decrease with increasing suspension stiffness values. The reason is that the suspension stiffness decides the force trans-
fer between chassis and road, which implies that there is a positive correlation between chassis-road coupling and suspension
stiffness. Therefore, the relation between cab/carriagechassis coupling and suspension stiffness is negative. However, the sus-
pension stiffness could only be used as a ne-tuning factor for their small sensitivities of inuence (all less than 1% when the
stiffness changes 20%), and the rear suspension stiffness has a higher effect (0.15%).
Fig. 10 shows the effect of cab mount stiffness on the coupling coecients of subsystem. For the coupling of cabchassis,
raising the stiffness value of front mount leads to a great increment of coupling effect. The stiffness of cab mounts inuence the
cabchassis coupling (23.8% for front mounts and 4.7% for rear mounts) in a much more sensitive way than the carriagechassis
coupling (both much less than 1%). Consequently, the cab mount stiffness could be seen as effective variables in the control of
cabchassis coupled vibration. Moreover, the stiffness of front cab mount is the more ecient one for adjusting the coupling
effect between cab and chassis.
It can be clearly seen from Fig. 11 that the effects of carriage mount stiffness on coupling coecient of cab is negative, mean-
while the impact is inuenced by the mount position. The inuence of front and rear mounts are larger than the middle ones.
JID: APM
(a)
(b)
Fig. 15. Relationship between vibration attenuation and coupling coecients, (a) cab, and (b) carriage.
The inuences are more complicated in the case of carriage coupling. In the vertical direction, the stiffness has a positive effect
on the substructure motion and coupling coecient. For the pitching and rolling motion, the effect of stiffness should be com-
prehensively considered combining with the mount position. Among four mounts positions (front, middle front, middle rear, and
rear), the stiffness of front mount and the middle rear mount (6.4% and 6.7%) are more sensitive than the middle front (3.8%) and
rear (3.6%) mounts for coupling effect of cabchassis subsystem; all the mount stiffness are insensitive (less than 0.2%) for the
case of carriagechassis subsystem.
JID: APM
4.7. Effect of substructure inertia parameter on coupling coecient
The inuences of inertial variables of cab, carriage, and chassis on the subsystem coupling coecients are shown in Fig. 1214.
It can be seen that the coupling coecient of cab (carriage) reduces when the cab (carriage) inertial variable grows. The reason
is that, the bigger value of cab/carriage inertial variable, the lower response under a specic excitation as well as the coupling
effect. While the coupling coecients of the cabchassis and the carriagechassis will increase together with the inertial variable
of chassis due to the coupling effect of chassis-road reducing with growing chassis inertia. The coupling effect of cabchassis is
more sensitive to the inertial properties than that of the carriagechassis subsystem. The moment of inertia of cab about the
lateral direction, the moment of inertia of carriage about the lateral direction, and the moment of inertial of chassis about the
longitudinal direction are the top three most sensitive variables, which change cab , 12.2%, 9.3%, and 8.0% respectively.
4.8. Relationship between coupling effect and vibration attenuation
The relationship between the vibration attenuation and subsystem coupling is investigated by tting the function, which
takes the subsystem coupling coecients as input and relative vibration attenuation coecients as output, using the BPNN
technique. Fig. 15(a) shows the relationship between the vibration reduction in cab and the two coupling coecients of cab
and carriage and Fig. 15(b) shows the correlation of carriage vibration reduction and the coupling effect of the two subsystem.
The mean square tting error is 0.0898 for the vibration attenuation coecient of cab and 0.0537 for the vibration attenuation
coecient of carriage. The vibration attenuation coecients of cab and carriage, cab and car , vary from 0 to 1, which represent
the suppressive level on vibration, and the closer to one the better. From the simulation results, we can conclude that the vibration
reduction for cab needs a low coupling between carriage and chassis while a high coupling of cab and chassis. Consequently, the
peak vibration attenuation coecient of cab appears at the point of (car = 0, cab = 1). Unfortunately, the vibration reduction
for carriage demands a fully reversed condition, which is the car = 1 and cab = 0. Obviously, it is impossible to get the best
vibration reduction results for both the cab and carriage in the same situation and a trade-off should be made. As a result, a
single objective optimization for either individual substructure might not be effective for the design or optimization of the whole
vehicle ride comfort. In order to get a proper balance or nd a compromise between vibration control of different substructures,
a multi-objective optimization method and the concerning weighting of objective are necessary for a reasonable design.
5. Conclusions
The dynamic model of a multi-supported system is developed and rewritten in the form of state space equations. The vibration
reduction of substructure is derived using state matrix and interaction matrix of the uncoupled system. Those two coecients are
able to quantify the degree of subsystem coupling and the vibration attenuation of substructure, so that the coupling correlation
can be revealed numerically and weighed in degree.
As an application of the proposed mathematical model and system, a 15-DOF vehicle model is established and described
as a cabcarriagechassis coupled vibration system. To validate the model, the predicted and measured responses of carriage
oor to random road excitation are compared at ve speeds and the results show a great agreement. In order to investigate the
effect of vehicle design parameters on the structure coupling, 52 design variables including mass, moment of inertia, stiffness,
damping, and joint location are dened with lower and upper limits. By using Latin hypercube sampling and quadratic regression,
the sensitivity analysis of design variables shows that joint location, stiffness and moment of inertia have the most signicant
inuence on the subsystem coupling.
In the case of cabchassis subsystem, the longitudinal distance from cab mount to cab CoG has the greatest inuence on
the coupling correlation and the stiffness of cab front mount is the second greatest affecting factor. The coupling effect of cab
chassis subsystem increases with the increment of the above two variables. While, the moment of inertia of cab about the
lateral direction has the most signicant effect among the negatively correlated variables, meantime, it has the third greatest
contribution to the cabchassis coupling in total. In the case of carriagechassis subsystem, the lateral distance from suspension
joint to chassis CoG has negative and the most signicant inuence on the coupling correlation. The next two variables are the
moment of inertia of chassis about the longitudinal direction and the longitudinal distance from carriage mount to carriage CoG,
and their effects are both positive.
In the assessment of vibration reduction of cab, a high coupling between cab and chassis and a synchronous low coupling
in carriagechassis are preferred to enhance the vibration attenuation coecient of cab. While, considering the carriage ride
comfort, a low coupling in cabchassis and a highly coupled carriagechassis are the best for vibration reduction. Consequently,
the lowest vibration responses of cab and carriage are not capable to achieve simultaneously due to the distinct trade-off of sub-
system coupling correlations. A multi-objective optimization will be employed in order to compromise the vibration reduction
in the coupled system in the future work.
Acknowledgments
This work has been supported by the National High Technology Research and Development Program of China
(2014AA041501).
JID: APM
References
[1] B.D. Van Deusen, Truck suspension system optimization, J. Terramech. 9 (1973) 83100.
[2] D.J. Cole, Fundamental issues in suspension design for heavy road vehicles, Vehicle Syst. Dyn. 35 (2001) 319360.
[3] M. Ieluzzi, P. Turco, M. Montiglio, Development of a heavy truck semi-active suspension control, Control Eng. Practice 14 (2006) 305312.
[4] M.M. Elmadany, Z. Abduljabbar, Design evaluation of advanced suspension systems for truck ride comfort, Comput. Struct. 36 (1990) 321331.
[5] D.J. Cole, D. Cebon, Truck suspension design to minimize road damage, proceedings of the institution of mechanical engineers, part D, J. Automobile Eng.
210 (1996) 95107.
[6] M. Valek, W. Kortm, Z. ika, L. Magdolen, O. Vaculn, Development of semi-active road-friendly truck suspensions, Control Eng. Practice 6 (1998) 735744.
[7] M. Valek, V. Stejskal, Z. ika, O. Vaculn, J. Kovanda, Dynamic model of truck for suspension control, Vehicle Syst. Dyn. 29 (1998) 496505.
[8] J.J. Kim, H.Y. Kim, Shape design of an engine mount by a method of parameter optimization, Comput. Struct. 65 (1997) 725731.
[9] J.S. Tao, G.R. Liu, K.Y. Lam, Design optimization of marine engine-mount system, J. Sound Vib. 235 (2000) 477494.
[10] D.H. Lee, W.S. Hwang, C.M. Kim, Design sensitivity analysis and optimization of an engine mount system using an frf-based substructuring method, J. Sound
Vib. 255 (2002) 383397.
[11] S. Arzanpour, N. Eslaminasab, B. Schubert, A. Narimani, M.F. Golnaraghi, A novel technique for frequency and time optimization of automotive engine mount
parameters, American Society of Mechanical Engineers, Dynamic Systems and Control Division (Publication) DSC, 2006.
[12] T. Ramachandran, K.P. Padmanaban, P. Nesamani, Modeling and analysis of IC engine rubber mount using nite element method and RSM, Procedia Eng. 38
(2012) 16831692.
[13] T. Ramachandran, K.P. Padmanaban, Minimization of IC engine rubber mount displacement using genetic algorithm, Int. J. Adv. Manuf. Technol. 67 (2013)
887898.
[14] N. Eslaminasab, S. Arzanpour, M.F. Golnaraghi, Optimal design of asymmetric passive and semi-active dampers, ASME Int. Mech. Eng. Congress Exposition
Proceedings (2008) 18411847.
[15] H. Mansour, S. Arzanpour, M.F. Golnaraghi, A.M. Parameswaran, Semi-active engine mount design using auxiliary magneto-rheological uid compliance
chamber, Vehicle Syst. Dyn. 49 (2011) 449462.
[16] Y. Nakaji, S. Satoh, T. Kimura, T. Hamabe, Y. Akatsu, H. Kawazoe, Development of an active control engine mount system, Vehicle Syst. Dyn. 32 (1999)
185198.
[17] F. Khameneifar, S. Arzanpour, M. Moallem, Piezo-actuated active decoupler hydraulic engine mount, in: Proceedings of ASME International Mechanical
Engineering Congress and Exposition (IMECE), 2010, pp. 195200.
[18] H. Mansour, S. Arzanpour, F. Golnaraghi, Design of a solenoid valve based active engine mount, JVC J. Vib. Control 18 (2012) 12211232.
[19] A.W. Foster, A heavy truck cab suspension for improved ride, SAE Technical Paper, 1978.
[20] H. Hoshino, T. Sakurai, K. Takahashi, Vibration reduction in the cabins of heavy-duty trucks using the theory of load transfer paths, JSAE rev. 24 (2003)
165171.
[21] K. Nakano, Combined type self-powered active vibration control of truck cabins, Vehicle Syst. Dyn. 41 (2004) 449473.
[22] J. Le, J. Cheng, W.-R. Harn, Simulation of Functional Stiffness for Generic Body Mount, SAE Technical Paper, 1998.
[23] P.M. Miller, J. Lee, R.K. Murthy, J. Cheng, Role of the body mount on the passenger compartment response of a frame/body structured vehicle in frontal crash,
SAE Technical Paper, 1998.
[24] J. Baum, J. Bennett, T. Carne, Truck ride improvement using analytical and optimization methods, SAE Technical Paper, 1977.
[25] G.-H. Lee, J.-H. Lim, G.-T. Kim, Improving ride quality on the cab suspension of a heavy duty truck, SAE Technical Paper, 1996.
[26] C. Ward, The application of a new cab mounting to address cab shake on the 2003 Chevrolet Kodiak and GMC TopKick, SAE Technical Paper, 2002.
[27] P. Xu, B. Bernardo, K. Tan, Optimal mounting design for cab vibration isolation, Int. J. Vehicle Des. 57 (2011) 292304.
[28] Y. Chen, T. Tyan, O. Faruque, Dynamic testing and cae modeling of body mount an application in the frontal impact analysis of a body-on-frame vehicle, SAE
Technical Paper, 2003.
[29] D. Kalsule, D. Hudson, Y. Yeola, J. Bohari, Structure borne noise and vibration reduction of a sports utility vehicle by body-mount dynamic stiffness opti-
mization, SAE Technical Paper, 2011.
[30] A. Preumont, Vibration control of active structures, Kluwer academic publishers Dordrecht 1997.
[31] B.D. Anderson, P. Agathoklis, E. Jury, M. Mansour, Stability and the matrix Lyapunov equation for discrete 2-dimensional systems, circuits and systems, IEEE
Trans. 33 (1986) 261267.
[32] H. Nakamura, K. Nakano, R. Oohashi, H. Hu, Y. Uchiyama, S. Kakihara, Road prole estimation from car body vibration using independent component analysis,
in: ASME 2012 5th Conference On Annual Dynamic Systems and Control Conference joint with the JSME 2012 11th Motion and Vibration, American Society
of Mechanical Engineers, 2012, pp. 563567.
[33] J.C. Helton, F.J. Davis, Latin hypercube sampling and the propagation of uncertainty in analyses of complex systems, Reliability Eng. Syst. Saf. 81 (2003)
2369.
[34] R.L. Iman, Latin hypercube sampling, Wiley Online Library, 2008.
[35] R. Hecht-Nielsen, Theory of the backpropagation neural network, Neural Networks, 1989. IJCNN, Conference on International Joint, IEEE (1989) 593605.
[36] J. Mor, The Levenberg-Marquardt algorithm: Implementation and theory, in: G.A. Watson (Ed.), Numerical Analysis, Springer Berlin Heidelberg, 1978,
pp. 105116.

A Mathematical Model For Coupled Vibration System of Road

Hochgeladen von

Dokumentinformationen

Originaltitel

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

A Mathematical Model For Coupled Vibration System of Road

Hochgeladen von

Copyright:

Verfügbare Formate

JID: APM

ARTICLE IN PRESS [m3Gsc;August 14, 2015;11:35]

Applied Mathematical Modelling 000 (2015) 119

Contents lists available at ScienceDirect

Applied Mathematical Modelling

A mathematical model for coupled vibration system of road

2 J. Zhang et al. / Applied Mathematical Modelling 000 (2015) 119

Superscript and subscript

J. Zhang et al. / Applied Mathematical Modelling 000 (2015) 119 3

Fig. 1. Schematic diagram of a multi-supported system.

2.1. Equation of motion

4 J. Zhang et al. / Applied Mathematical Modelling 000 (2015) 119

2.2. State space description

J. Zhang et al. / Applied Mathematical Modelling 000 (2015) 119 5

Then, the state equation of the coupled vibration system is given by

2.3. Dynamic characteristic

Using the ratio of two-norm of H, the vibration attenuation coecient is dened as

DT H (D)+H (D)D + I = 0, (29)

so that the expression of i is given by

6 J. Zhang et al. / Applied Mathematical Modelling 000 (2015) 119

Fig. 2. Schematic diagram of the vehicle model.

2.4. Vehicle model

C cha = E susT sus cabT cab carT car

K cha = E susT sus cabT cab carT car

J. Zhang et al. / Applied Mathematical Modelling 000 (2015) 119 7

Variable Denition Unit Range

where l is the design parameter which is listed in Table 1.

8 J. Zhang et al. / Applied Mathematical Modelling 000 (2015) 119

K cab = E chaT cha

K car = E chaT cha

where C cab cha cab cha

J. Zhang et al. / Applied Mathematical Modelling 000 (2015) 119 9

The uncoupling system matrix can be written as

The coupling characteristic matrixes can be obtained as

(5) Transfer function

where is the Laplace operator [32]. Extending Eq. (63) gives

where L() is Laplace transformation [32].

10 J. Zhang et al. / Applied Mathematical Modelling 000 (2015) 119

Fig. 3. Schematic diagram of loading position of force on chassis.

Fig. 4. Comparison of calculated and measured vibration responses at carriage oor.

4. Analysis and result

4.1. Model validation

J. Zhang et al. / Applied Mathematical Modelling 000 (2015) 119 11

Order Cab Carriage

Design variable Contribution % Design variable Contribution %

1 l11 15.49 l4 15.60

Fig. 5. Effects of position variable of cab mount on coupling coecients.

4.2. Latin hypercube sampling

4.3. Back propagation neural networks

12 J. Zhang et al. / Applied Mathematical Modelling 000 (2015) 119

4.4. Contribution analysis

J. Zhang et al. / Applied Mathematical Modelling 000 (2015) 119 13

Fig. 7. Effects of position variable of suspension on coupling coecients.

Fig. 8. Effects of position variable of substructure centroid on coupling coecients.

4.5. Effect of suspension/mount position on coupling coecient

14 J. Zhang et al. / Applied Mathematical Modelling 000 (2015) 119

Fig. 9. Effects of stiffness variable of suspension on coupling coecients.

Fig. 10. Effects of stiffness variable of cab mount on coupling coecients.

J. Zhang et al. / Applied Mathematical Modelling 000 (2015) 119 15

Fig. 11. Effects of stiffness variable of carriage mount on coupling coecients.

Fig. 12. Effects of inertia variable of cab on coupling coecients.

16 J. Zhang et al. / Applied Mathematical Modelling 000 (2015) 119

Fig. 13. Effects of inertia variable of carriage on coupling coecients.