Sunteți pe pagina 1din 7

This article has been accepted for inclusion in a future issue of this journal.

Content is final as presented, with the exception of pagination.


IEEE TRANSACTIONS ON CONTROL SYSTEMS TECHNOLOGY

Performance Benefits of Using Inerter in Semiactive Suspensions


Michael Z. Q. Chen, Yinlong Hu, Chanying Li, and Guanrong Chen

Abstract This brief investigates the performance benefits of


using inerter in semiactive suspensions. A novel structure for
semiactive suspensions with inerter, which consists of a passive
part and a semiactive part, is proposed. Six semiactive suspension
struts, each of which employs an inerter in the passive part
and a semiactive damper in the semiactive part, are introduced.
Two suboptimal control laws named clipped optimal control and
steepest gradient control laws are derived to control the damping
coefficient in the semiactive part. Extensive simulations with
respect to different choices of weighting factors and suspension
static stiffnesses are conducted based on both a quarter car
model and a full car model. The results show that, compared
with the conventional semiactive suspension strut, the overall
suspension performance can be significantly improved using
inerters, including ride comfort, suspension deflection, and road
holding. Comparative studies between these two suboptimal
control laws and between these semiactive struts are also carried
out to facilitate future practical application of the proposed
semiactive suspensions with inerter.
Index Terms Full car model, inerter, quarter car model,
semiactive suspension, suboptimal control.

I. I NTRODUCTION

NERTER is a two-terminal mechanical device with the


property that the applied force at the two terminals is proportional to the relative acceleration between them [1], whose
symbolic representation is shown in Fig. 1. The inerter expands
the class of mechanical realizations of complex impedances
compared with the ones using only springs and dampers,
and has been applied to various mechanical systems, such as
vehicle suspensions [2][6]. It has also rekindled interest in
passive network synthesis [7][14].
Semiactive suspension has attracted much attention because
of its low energy consumption compared with the active
one [15][17] and its high performance compared with
Manuscript received May 27, 2014; accepted October 19, 2014. Manuscript
received in final form October 20, 2014. This work was supported in part by
the National Natural Science Foundation of China under Grant 61374053
and Grant 61203067, in part by the Hong Kong University Committee
on Research and Conference Grants under Grant 201211159112, and in
part by the Research Grants Council, Hong Kong, through the General
Research Fund under Grant CityU 1120/14. Recommended by Associate
Editor S. M. Savaresi.(Corresponding author: Michael Z. Q. Chen.)
M. Z. Q. Chen is with the Department of Mechanical Engineering, University of Hong Kong, Hong Kong, and also with the Shenzhen Institute
of Research and Innovation, University of Hong Kong, Hong Kong (e-mail:
mzqchen@hku.hk).
Y. Hu is with the School of Automation, Nanjing University of Science and
Technology, Nanjing 210044, China (e-mail: yinlong.h@gmail.com).
C. Li is with the Academy of Mathematics and Systems Sciences, Chinese
Academy of Sciences, Beijing 100190, China (e-mail: cyli@amss.ac.cn).
G. Chen is with the Department of Electronic Engineering, City University
of Hong Kong, Hong Kong (e-mail: gchen@ee.cityu.edu.hk).
Color versions of one or more of the figures in this paper are available
online at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TCST.2014.2364954

Fig. 1. Symbolic representation of inerter with F = b(v2 v1 ), where b is


called inertance, which has units of kilograms [1].

the passive one. The conventional semiactive suspension


configuration, that is, a spring in parallel with a variable
shock absorber (may contain a passive damper), has been
extensively investigated with a large number of meaningful
results obtained [18][23].
This brief proceeds to applying inerter in semiactive
suspensions, showing its significant performance benefits as
well. These are demonstrated by proposing a novel structure for semiactive suspensions employing inerters, which
consists of a passive part and a semiactive part. Six passive networks, each of which employs an inerter, constitute
the passive part, whereas the semiactive part incorporates
a semiactive damper. Two suboptimal control laws named
clipped optimal control (COC) law and steepest gradient
control (SGC) law [21] are employed to control the semiactive
damper, and extensive simulations based on both a quarter
car model and full car model are conducted to demonstrate
the significant improvements of using inerter in semiactive
suspensions. All these constitute the main contributions of
this brief.
The organization of this brief is as follows. In Section II, the
proposed semiactive suspensions with inerter are introduced
based on a quarter car model and the semiactive suspension
control problem is formulated. In Section III, two suboptimal
control laws for these six semiactive suspension struts are
derived, and in Section IV, numerical simulation in terms of a
quarter car model is carried out. A study of the proposed semiactive suspensions in a full car model is reported in Section V.
Conclusions are drawn in Section VI.
The notation used throughout this brief is standard. I and 0
are used to denote the identity matrix and zero matrix of
appropriate dimensions, respectively, and diag are used to
denote the diagonal matrix.
II. Q UARTER C AR M ODEL AND
P ROBLEM F ORMULATION
A. Quarter Car Model
Consider the quarter car model shown in Fig. 2, which
consists of the sprung mass m s , unsprung mass m u , and tire
vertical stiffness kt [20]. The suspension system consists of

1063-6536 2014 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.

This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
2

IEEE TRANSACTIONS ON CONTROL SYSTEMS TECHNOLOGY

TABLE I
W (s) FOR E ACH C ONFIGURATION IN F IG . 3, W HERE s
D ENOTES THE L APLACE VARIABLE

A state space model for the semiactive quarter car system


can easily be obtained as
Fig. 2.

Semiactive quarter car model.

xm = Am x m + Bm Fp + Bm Fd + Bmr z r

(1)

where x m = [z s z u z s z u ]T , Fp denotes the force generated


by the networks W (s), and




0
I
0
Am =
, Bm =
M 1 K M 1 Cv0
M 1 E


0
Bmr =
, M = diag{m s , m u }, E = [1 1]T
M 1 K t




cv0
ks
ks
c
, K=
.
K t = [0 kt ]T, Cv0 = v0
cv0 cv0
ks ks +kt
Define F p = sW (s)(z s z u ), where F p , z s , and z u denote
the Laplace transforms of Fp , z s , and z u , respectively. A state
space model for the passive part W (s) can be obtained as
x p = A p x p + B p (z s z u )
Fp = C p x p + D p (z s z u )

(2)
(3)

where z s z u is the input and Fp is the output. Combining


(1)(3), the semiactive quarter car system can be rewritten as
Fig. 3. Configurations as W (s) of the suspensions in Fig. 2. (a) C1. (b) C2.
(c) C3. (d) C4. (e) C5. (f) C6. (g) C7.

two parts: 1) the passive part and 2) semiactive part. The passive part contains a spring in parallel with a passive network
shown in Fig. 3, where the admittance of the passive part is
Yi (s) = (ks /s) + Wi (s), i = 1, . . . , 7, and Wi (s) is shown
in Table I. Here, admittance is defined as the ratio of force
to velocity according to the force-current analogy [1], [7].
The semiactive part involves a variable shock absorber, such
as electrohydraulic dampers (EH dampers), magnetorheological dampers (MR dampers), or electrorheological dampers
(ER dampers) [19].
Denote cv as the controllable damping coefficient, which
can be adjusted by modifying the electronic valves for
EH dampers or changing the physical properties of fluids
for MR dampers and ER dampers (see [19, Ch. 2] for
details). The force generated by the semiactive part Fd can
be represented as Fd = cv (z s z u ), where cv [cmin , cmax ].
In this brief, a nominal damping coefficient cv0 is considered
for the semiactive damper so that a stable open-loop plant
can be obtained, which means that the semiactive damping
coefficient cv is adjusted based on cv0 . In this way, the lower
and upper bounds for cv become cmin cv0 and cmax cv0 ,
respectively.

x g = A g x g + Bg Fd + Bgr z r

(4)

where x g = [x mT x Tp ]T and




A m + Bm D p L Bm C p
Bm
Ag =
, Bg =
Bp L
Ap
0


Bmr
Bgr =
, L = [0 0 1 1].
0
B. Road Model
Typically, random road profiles can be described using a
power spectral density (PSD) function in the form of
 

() = (0 )
0
where  is the wave number with the unit [rad/m] and
0 = (0 ) in [m2 /(rad/m)] is the value of the PSD at
0 = 1 rad/m. In addition, denotes the waviness, where
= 2 for most of the road surfaces. To realize such a PSD
function in the time domain, the shaping filter method [25]
can be employed as follows:
zr (t) = V z r (t) + w(t)

(5)

where the white noise process w(t) with the spectral density
w = 2V 2 and V is the vehicle forward speed.

This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
CHEN et al.: PERFORMANCE BENEFITS OF USING INERTER IN SEMIACTIVE SUSPENSIONS

TABLE II
ROAD D ESCRIPTION C LASSIFIED BY ISO 8608 [24], [25]

where P is the solution of the following algebraic Riccati


equation (ARE):
A P + P A (P B + N)R 1 (B T P + N T ) + Q = 0.

As suggested by the International Organization for Standardization (ISO) [24], typical road profiles based on the
value of 0 can be classified as Class AClass E, as shown
in Table II. The values of and in the linear filter
that correspond to the ISO standard road profiles are shown
in Table II (see [25] for details).
C. Augmented Model and Problem Formulation
Combining (4) and (5), and denoting x = [x gT , z r ]T , one
can obtain an augmented model as
x = Ax + B Fd + Bw w
where

A=

Ag
0


Bgr
,
V


B=


Bg
,
0

(6)
 
0
.
Bw =
1

y = [z s (z s z u ) (z u z r )]T = C x + D Fd

A(3, :)
C = 1 1 0 0 0 0 ,
0 1 0 0 0 1

where
Fdmin = (cmin cv0 )|z s z u |,

Fdmax = (cmax cv0 )|z s z u |.

where P is the solution of the Lyapunov equation


A T P + P A = Q.

Bm (3, :)

0
D=
0

and A(3, :) and Bm (3, :) denote the third row of A and Bm ,


respectively.
To deal with these three performance requirements simultaneously, a quadratic performance index is defined as

T
y T y + Fd2 r dt
(7)
J = lim
T 0

where  contains the weighting factors among sprung mass


acceleration, suspension deflection, and tire deflection, and r is
a weighting factor to be determined by the designer.
The performance index (7) can be rewritten as

T
J = lim
x T Qx + 2x T N Fd + FdT R Fd dt
(8)
T 0

where Q = C T C, N = C T D and R = D T D + r .


III. S UBOPTIMAL C ONTROLLER D ESIGN
If ignoring the constraint on the semiactive damping, the
aforementioned problem is a linear quadratic regulation problem and an optimal controller can be derived as
Fd = R 1 (B T P + N T )x

As shown in [21], the optimal solution for a quarter car


semiactive suspension involves time-varying Riccati equations
and no analytical solution can be obtained. However,
suboptimal control laws based on, such as the clipped optimal
method and the steepest gradient method, are fine engineering
approximations in practice, which are derived, respectively,
in the following.
Proposition 1: The COC law is

Fdmin , Fd < Fdmin


(11)
Fd = Fd ,
Fdmin Fd Fdmax

Fdmax , Fd > Fdmax

As shown in [21], the performance index (8) is related to


the optimal passive suspension as


 T

2x N Fd + FdT R Fd dt
Jsemi = Jpassive +

The controlled output y involving the sprung mass acceleration, suspension deflection, and tire deflection is defined as

where

(10)

(9)

(12)

The SGC law is derived as follows.


Proposition 2: The SGC law can be derived using (11)
where the P matrix in (9) is derived by (12) instead
of (10).
Note that the COC law and SGC law share the same
structure, where the only difference is the procedure to
determine P. The COC law is equivalent to making the semiactive suspension approximate the active suspension, whereas
the SGC law is equivalent to increasing the improvement of
the semiactive suspension compared with the passive one [21].
Since both of these two control laws heavily depend on the
values of the weight coefficients, it is hard to say, which
is better for a general semiactive suspension system [22].
In this brief, both of them are employed to demonstrate the
benefits of the inerter for the considered semiactive suspension
systems.
IV. Q UARTER C AR M ODEL S IMULATION
The vehicle parameters are taken from [2] with
m s = 250 kg, m u = 35 kg, kt = 150 kN/m, and the
suspension static stiffness ks is selected from 10 to 120 kN/m.
The vehicle forward speed V is assumed to be 30 m/s and
the nominal value of the semiactive damper is chosen as
cv0 = 1500 Ns/m. In addition, r in (7) is chosen as 0 and
the bounds of semiactive damping coefficient are chosen as
cmin = 0 Ns/m and cmax = 3000 Ns/m.

This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
4

IEEE TRANSACTIONS ON CONTROL SYSTEMS TECHNOLOGY

A. Quantitative Performance Indexes


To quantitatively compare the performances among these
configurations shown in Fig. 3, the performance measures ,
acc , sus , and rhd representing the overall, ride comfort,
suspension deflection, and road holding performances,
respectively, are defined as the root mean square (rms) values
of y, z s , z s z u , and z u z r , respectively.
The weighting factor  is chosen as


1
2
1
(13)
,
,
 = diag
norm 2 norm 2 norm 2
acc
sus
rhd

TABLE III
PARAMETERS IN PASSIVE PART W HEN ks = 80 kN/m FOR Q UARTER C AR
M ODEL (k S A RE IN kN/m, c S A RE IN kNs/m, b S A RE IN kg)

norm , norm , and norm are the normalized values


where acc
sus
rhd
of acc , sus , and rhd , respectively. In this brief, based
on the open-loop system (passive system) with the nominal damping coefficients cv0 = 1500 Ns/m, one obtains
norm = 0.9962, norm = 0.0058, and norm = 0.0028. Note
acc
sus
rhd
that for the above choice of , one has

2 =

2
2
2
rhd
acc
sus
+

1
2
norm 2
norm 2
norm 2
acc
sus
rhd

where 1 and 2 are weighting factors between ride comfort,


suspension deflection, and road holding.
In this brief, three groups of 1 and 2 values are selected as

1 = 0.5, 2 = 0.5 :

Preference of ride comfort over road holding

1 = 0.5, 2 = 1 :
Equal preference of ride comfort and road holding

1 = 0.5, 2 = 1.5 :

Preference of road holding over ride comfort.

Fig. 4.
Comparison of the overall performance when 1 = 0.5,
2 = 0.5 under the COC. Left: quantitative values. Right: percentage
improvement over C1.

B. Optimization of the Passive Part Parameters


The NelderMead simplex method is employed with various
starting points for each configuration to obtain the optimal
values of the component coefficients (a similar procedure can
be found in [2] and [4]), and a set of optimal parameters
are obtained with respect to different static stiffness and
different choices of 1 and 2 . The detailed values of these
parameters at ks = 80 kN/m are shown in Table III. Since
C4 and C7 reduce to C3 and C6, respectively, they are not
shown in Table III. Note that the nominal damping cv0 has
been considered in optimizing the passive part, which means
that a parallel damper with damping coefficient cv0 is added
to each configuration when acting as passive suspension to
obtain the optimal parameters.

Fig. 5. Comparison of the ride comfort, suspension deflection, and road


holding when 1 = 0.5, 2 = 0.5 under the COC.

C. Simulation Results
Numerical simulations under both the COC and SGC are
conducted, where the simulation time is chosen as 20 s and
the average road profile (Class C road) is employed. Note
that similar results are obtained from these two suboptimal
control laws. For brevity, only the results under the COC
are shown in Figs. 49. Since the relaxation spring kb in
C4 and C7 provides no improvement for ride comfort [2], [5],
C4 and C7 reduce to C3 and C6, respectively, when 1 = 0.5
and 2 = 0.5.

Fig. 6.
Comparison of the overall performance when 1 = 0.5,
2 = 1 under the COC. Left: quantitative values. Right: percentage improvement over C1.

From Figs. 4, 6, and 8, it is observed that for all the


selected 1 and 2 , the configurations with inerter perform
better than the conventional semiactive strut C1, where over
10% improvements have been obtained. It is also shown in
Figs. 4, 6, and 8 that the series arrangements of inerter

This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
CHEN et al.: PERFORMANCE BENEFITS OF USING INERTER IN SEMIACTIVE SUSPENSIONS

Fig. 7. Comparison of the ride comfort, suspension deflection, and road


holding when 1 = 0.5, 2 = 1 under the COC.

Fig. 10. Percentages of improvement of the SGC over the COC. Positive
sign means improvement, whereas negative sign means degradation.

Fig. 8.
Comparison of the overall performance when 1 = 0.5,
2 = 1.5 under the COC. Left: quantitative values. Right: percentage
improvement over C1.

Fig. 11.

Semiactive full car model.

V. A PPLICATION TO A F ULL C AR M ODEL

Fig. 9. Comparison of the ride comfort, suspension deflection, and road


holding when 1 = 0.5, 2 = 1.5 under the COC.

(C3C7) provide more improvement than the parallel arrangement C2. From Figs. 5, 7, and 9, it is seen that both ride
comfort and road holding are simultaneously improved by the
series arrangements C3C7. Meanwhile, suspension deflection
is degraded, which is consistent with the results in [5] on
the analysis of passive suspensions with inerter. However, the
improvement of the ride comfort by the parallel arrangement
C2 is obtained by sacrificing both suspension deflection and
road holding, which is a potential drawback of the parallel
arrangement of inerter.
A comparison between the COC and SGC is also conducted
and the percentages of improvement of the SGC over the
COC are shown in Fig. 10, where we see that for the selected
1 and 2 , these two suboptimal control laws do not differ
from each other by too much (only gaps of at most 1.5% are
shown in Fig. 10). It is also observed that for the considered
vehicle model, the COC tends to perform better than the
SGC in the low static stiffness range (1030 kN/m), whereas
the SGC tends to do better in the high static stiffness range
(70120 kN/m).

In this section, we proceed to applying the proposed semiactive suspensions with inerter to a seven degree of freedom full
car model, as shown in Fig. 11, whereas a detailed description
of this model can be found in [2]. The following subscripts
are used: 1) f r ; 2) f l; 3) rr ; and 4) rl denote the front-right,
front-left, rear-right, and rear-left, respectively. The proposed
semiactive suspensions are equipped at each corner of the full
car model, and it is assumed that the passive part at each corner
of the vehicle possesses the same structure, but the values of
the components coefficients can be different, which will be
determined by numerical optimization.
By assuming that the angles of the sprung mass in the
pitch and roll directions are small enough, and using the
method of modeling the passive parts in the quarter car case,
a state-space model integrating the full car model, the passive
parts, and a first-order road model can be obtained as follows:
x = Ax + B Fd + Bw w
where x = [x sT , x uT , x Tp , xrT ]T , x s
z u f l , z urr , z url ]T , x p , and xr are

, , ]T ,

(14)

= [z s
x u = [z u f r ,
states of the passive part
model and road model, respectively. Fd contains the forces
generated by the semiactive damper at each corner.
The road input considered in this full car case is a parallel
track, which can be obtained by passing different white noise
processes wl (t) and wr (t) through the same linear filter (5)
and ignoring the correlation between the left and right road
excitations. The Pad approximation method is employed to
approximate the time delay between the front wheels and the
rear wheels.

This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
6

IEEE TRANSACTIONS ON CONTROL SYSTEMS TECHNOLOGY

TABLE IV
PARAMETERS IN PASSIVE PART W HEN k f = kr = 80 kN/m FOR THE F ULL

The controlled outputs are defined as


T

y = xsT (x 1 x u )T (x u Z r )T

C AR M ODEL (k S A RE IN kN/m, c S A RE IN kNs/m, b S A RE IN kg)

where xs , (x 1 x u ), and (x u Z r ) represent the sprung


mass accelerations, suspension deflections, and tire deflections,
respectively, and
x 1 = [z s f r , z s f l , z srr , z srl ]T
= E T xs
Z r = [z r f r , z r f l , z rrr , z rrl ]T

1
1
1
1
E = l f
l f
lr
lr .
tf
t f
tr
tr
Similarly, a performance index is defined as

T
y T y + FdT Fd dt
J = lim
T 0

T
= lim
x T Qx + 2x N Fd + FdT R Fd dt
T 0
C T C, N

(15)

= C T D, and R = D T D+ , where


where Q =
is a weighting factor determined by the designer, and the
weighting factor  is chosen as  = diag{1 , 2 , 3 }, with


1
1
1
1 = diag
,
,
norm 2 norm 2 norm 2
acc
acc
acc


1
1
1
1
2 = 1 diag
,
,
,
norm 2 norm 2 norm 2
norm 2 sus
sus
sus

 sus
1
1
1
1
3 = 2 diag
,
,
,
norm 2 norm 2 norm 2 norm 2
rhd
rhd
rhd
rhd
norm , norm , and norm are the normalized values of
where acc
sus
rhd
acc , sus , and rhd , respectively. acc , sus , and rhd are the
rms values of xs , (x 1 x u ) and x u z r , respectively, which
quantitatively represent ride comfort, suspension deflection,
and road holding, respectively. In this way, one has

2 =

2
2
2
rhd
sus
acc
+ 1 norm
+ 2 norm
2
2
norm
acc
sus
rhd 2

where 1 and 2 are weighting factors between ride comfort,


suspension deflection, and road holding.
A. Suboptimal Control Laws
Similar to the quarter car case, two suboptimal control laws
can be derived for the considered full car model.
Proposition 3: The COC law is

< F
Fdi_ min , Fdi
di_ min

F
Fdi_min Fdi
(16)
Fdi = Fdi ,
di_ max

Fdi_ max , Fdi > Fdi_ max


where i denotes f r , f l, rr , and rl, separately
Fdi_ min = (cmin cv0i )|x1i xui |
Fdi_ max = (cmax cv0i )|x1i xui |
Fd = R 1 (B T P + N T )x

and P is the solution of the following ARE:


A P + P A (P B + N)R 1 (B T P + N T ) + Q = 0.

(17)

Fig. 12.
Comparison of the overall performance when 1 = 0.5,
2 = 1 under the COC. Left: quantitative values. Right: percentage improvement over C1.

Proposition 4: The SGC law can be derived using (16)


but the P matrix in (17) is derived by solving the following
Lyapunov function A T P + P A = Q.
B. Full Car Model Simulation
The following parameters taken from [2] are employed:
1) m s = 1600 kg; 2) I = 1000 kgm2 ; 3) I = 450 kgm2 ;
4) t f = 0.75 m; 5) tr = 0.75 m; 6) l f = 1.15 m;
7) lr = 1.35 m; 8) m f = 50 kg; 9) m r = 50 kg; 10) kt f =
250 kN/m; and 11) ktr = 250 kN/m. The vehicle forward
speed V is assumed to be 30 m/s and the nominal values of the
semiactive damper are chosen as cv0i = 1500 Ns/m, i denotes
f r , f l, rr , and rl, separately, and the bounds of semiactive
damping coefficient are chosen as cmin = 0 Ns/m and
cmax = 3000 Ns/m. Since similar results are obtained for
different choices of 1 and 2 in the quarter car case,
1 = 0.5 and 2 = 1 are selected in the full car model
simulation. The average road (Class C road) is employed
and the simulation time is 20 s. in (15) is chosen as diag{1e6 , 1e6 , 1e6 , 1e6 }. The normalized values are obtained from the open-loop simulation as
norm = 1.3837, norm = 0.0137, and norm = 0.0060.
acc
sus
rhd
Note that in the quarter car simulation, C7 always reduces
to C6 for most of static stiffness, and C6 performs slightly
better than C5 due to the employment of an extracentering
spring. Hence, in the full car case, the parallel arrangement C2
and the series arrangement C6 are considered, which are
compared with the conventional strut C1. First of all, the
optimal parameters of the passive parts at each corner are
obtained using the NelderMead simplex method. Similar
to the quarter car case, a set of optimal parameters with
respect to different static stiffness are obtained and the detailed
parameters at 80 kN/m are summarized in Table IV.
Then, a semiactive damper is equipped at each corner,
which is controlled by the proposed suboptimal control laws.

This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
CHEN et al.: PERFORMANCE BENEFITS OF USING INERTER IN SEMIACTIVE SUSPENSIONS

which is expected to find some applications in the near


future.
R EFERENCES

Fig. 13. Comparison of the ride comfort, suspension deflection, and road
holding when 1 = 0.5, 2 = 1 under the COC.

Fig. 14. Comparison between the SGC and the COC. Positive number means
SGC better than COC, whereas negative number means COC better than SGC.

As shown in Figs. 12 and 13, the simulation results confirm


the findings in the quarter car simulation that the proposed
semiactive suspensions with inerter indeed improve the overall performance compared with the conventional semiactive
suspension C1, where almost 7% improvement is shown
in Fig. 12, and both ride comfort and road holding can be
improved by the series arrangement C6. However, as shown
in Fig. 14, for the considered full car model, the COC
consistently performs better than the SGC except C2 at high
static stiffness, which is different from the quarter car case.
This confirms the conclusion that the performance of these
suboptimal control laws heavily depends on the selected
parameters [22].
VI. C ONCLUSION
In this brief, performance benefits of using inerter in semiactive suspensions have been demonstrated by proposing a
novel structure for semiactive suspension with inerter, which
consists of a passive part and a semiactive part. Six different
semiactive suspension struts, each of which employs an inerter
in the passive part and a semiactive damper in the semiactive
part, were introduced. Two suboptimal control laws, that is,
the COC and the SGC laws, were derived to control the semiactive damping coefficients. An augmented state-space model
integrating the vehicle model, the suspension model, and a
first-order road model was established for both a quarter
car model and a full car model. Extensive simulations
were conducted, showing that the proposed semiactive struts
can significantly improve the overall suspension performance, including ride comfort, suspension deflection, and road
holding. For example, >10% improvements over the conventional semiactive suspension without inerter were observed for
the quarter car model. Moreover, comparisons between these
two suboptimal control laws and between these semiactive
suspension struts, were conducted to facilitate the practical
implementation of the proposed semiactive suspension struts,

[1] M. C. Smith, Synthesis of mechanical networks: The inerter, IEEE


Trans. Autom. Control, vol. 47, no. 10, pp. 16481662, Oct. 2002.
[2] M. C. Smith and F.-C. Wang, Performance benefits in passive vehicle
suspensions employing inerters, Veh. Syst. Dyn., Int. J. Veh. Mech.
Mobility, vol. 42, no. 4, pp. 235257, 2004.
[3] F.-C. Wang and H.-A. Chan, Vehicle suspensions with a mechatronic
network strut, Veh. Syst. Dyn., Int. J. Veh. Mech. Mobility, vol. 49, no. 5,
pp. 811830, 2011.
[4] M. Z. Q. Chen, Y. Hu, and B. Du, Suspension performance with one
damper and one inerter, in Proc. 24th Chin. Control Decision Conf.,
Taiyuan, China, May 2012, pp. 35343539.
[5] Y. Hu, M. Z. Q. Chen, and Z. Shu, Passive vehicle suspensions
employing inerters with multiple performance requirements, J. Sound
Vibrat., vol. 333, no. 8, pp. 22122225, 2014.
[6] Y. Hu, C. Li, and M. Z. Q. Chen, Optimal control for semi-active
suspension with inerter, in Proc. 31st Chin. Control Conf., Hefei, China,
Jul. 2012, pp. 23012306.
[7] M. Z. Q. Chen, Passive network synthesis of restricted complexity,
Ph.D. dissertation, Dept. Eng., Univ. Cambridge, Cambridge, U.K.,
2007.
[8] M. Z. Q. Chen and M. C. Smith, Electrical and mechanical passive
network synthesis, in Recent Advances in Learning and Control,
vol. 371. New York, NY, USA: Springer-Verlag, 2008, pp. 3550.
[9] M. Z. Q. Chen and M. C. Smith, A note on tests for positive-real
functions, IEEE Trans. Autom. Control, vol. 54, no. 2, pp. 390393,
Feb. 2009.
[10] M. Z. Q. Chen and M. C. Smith, Restricted complexity network realizations for passive mechanical control, IEEE Trans. Autom. Control,
vol. 54, no. 10, pp. 22902301, Oct. 2009.
[11] K. Wang and M. Z. Q. Chen, Generalized seriesparallel R LC synthesis without minimization for biquadratic impedances, IEEE Trans.
Circuits Syst. II, Exp. Briefs, vol. 59, no. 11, pp. 766770, Nov. 2012.
[12] M. Z. Q. Chen, K. Wang, Z. Shu, and C. Li, Realizations of a special
class of admittances with strictly lower complexity than canonical
forms, IEEE Trans. Circuits Syst. I, Reg. Papers, vol. 60, no. 9,
pp. 24652473, Sep. 2013.
[13] M. Z. Q. Chen, K. Wang, Y. Zou, and J. Lam, Realization of a special
class of admittances with one damper and one inerter for mechanical
control, IEEE Trans. Autom. Control, vol. 58, no. 7, pp. 18411846,
Jul. 2013.
[14] K. Wang, M. Z. Q. Chen, and Y. Hu, Synthesis of biquadratic
impedances with at most four passive elements, J. Franklin Inst.,
vol. 351, no. 3, pp. 12511267, 2014.
[15] H. Du and N. Zhang, H control of active vehicle suspensions with
actuator time delay, J. Sound Vibrat., vol. 301, nos. 12, pp. 236252,
2007.
[16] H. Du and N. Zhang, Fuzzy control for nonlinear uncertain electrohydraulic active suspensions with input constraint, IEEE Trans. Fuzzy
Syst., vol. 17, no. 2, pp. 343356, Apr. 2009.
[17] Y. Hu, M. Z. Q. Chen, and Z. Hou, Multiplexed model predictive
control for active vehicle suspensions, Int. J. Control, to be published.
[18] H. Du, K. Y. Sze, and J. Lam, Semi-active H control of vehicle suspension with magneto-rheological dampers, J. Sound Vibrat., vol. 283,
nos. 35, pp. 981996, 2005.
[19] S. M. Savaresi, C. Poussot-Vassal, C. Spelta, O. Sename, and L. Dugard,
Semi-Active Suspension Control Design for Vehicles. Amsterdam,
The Netherlands: Elsevier, 2010.
[20] C. Poussot-Vassal, S. M. Savaresi, C. Spelta, O. Sename, and
L. Dugard, A methodology for optimal semi-active suspension systems
performance evaluation, in Proc. 49th IEEE Conf. Decision Control,
Atlanta, GA, USA, Dec. 2010, pp. 28922897.
[21] H. E. Tseng and J. K. Hedrick, Semi-active control lawsOptimal and
sub-optimal, Veh. Syst. Dyn., Int. J. Veh. Mech. Mobility, vol. 23, no. 1,
pp. 545569, 1994.
[22] M. Rienks, A comparison of two control laws for semi-active
suspensions, Tech. Univ. Eindhoven, Eindhoven, The Netherlands,
WFW Rep. 94.036, Mar. 1994.
[23] S. M. Savaresi and C. Spelta, A single-sensor control strategy for semiactive suspensions, IEEE Trans. Control Syst. Technol., vol. 17, no. 1,
pp. 143152, Jan. 2009.
[24] Mechanical VibrationRoad Surface ProfilesReporting of Measured
Data, ISO Standard 8608:1995-09-01, 1995.
[25] F. Tyan, Y.-F. Hong, S.-H. Tu, and W. S. Jeng, Generation of random
road profiles, J. Adv. Eng., vol. 4, no. 2, pp. 13731378, 2009.

S-ar putea să vă placă și