You are on page 1of 7

Int. J. Dynam.

Control
DOI 10.1007/s40435-014-0128-z

Fuzzy control of passenger ride performance using MR shock


absorber suspension in quarter car model
Devdutt · M. L. Aggarwal

Received: 1 April 2014 / Revised: 7 September 2014 / Accepted: 9 September 2014


© Springer-Verlag Berlin Heidelberg 2014

Abstract This paper presents vibration control perfor- ride comfort and safety of travelling passengers based on the
mance of quarter car model with three-degrees-of-freedom assembled parts. Passive suspension system delivers limited
having Magneto- Rheological (MR) suspension system. performance in vibration control applications due to uncon-
Experimental work is performed on an MR shock absorber trollable shock absorber part. While, high cost and complex
prototype under various cyclic excitation conditions. A poly- technology of assembled parts in active suspension system
nomial model is selected to characterize the test results of put forward a major hindrance for their commercial use.
MR shock absorber. The designed forward fuzzy logic con- A reliable choice compared to passive and active suspen-
troller (FFLC) and inverse fuzzy logic controller (IFLC) are sion technology is semi-active suspension control [1]. Semi-
assembled in the secondary suspension system of quarter car active suspension system can provide desirable performance
model. The response plots in time domain due to bump road with less power as well as cost requirement compared to pas-
disturbance related to passenger seat are obtained for uncon- sive and active suspension system technology. Research and
trolled and controlled quarter car models. Simulation results development in the field of externally controllable magneto-
are compared for selection of best option which can provide rheological (MR) shock absorbers has made semi-active sus-
maximum ride comfort to travelling passengers. Simulation pension technology practically feasible.
results demonstrate that semi-active quarter car system pro- Different researchers have studied the practical applica-
vides improved overall performance in terms of passenger tion of MR damper in semi-active suspension system. Sassi
ride comfort and safety compared to uncontrolled system. et al. [2] proposed theoretical and experimental work related
to the optimization of the dynamic response of a MR damper.
Keywords Magneto-rheological shock absorber · Semi- The performance of the developed damper was tested under
active suspension system · Quarter car model · Three degrees harmonic excitation to verify the requirements related to
of freedom · Fuzzy logic control comfort and stability for semi-active suspension system.
Nguyen et al. [3] studied an optimal design of a MR shock
absorber taking the damping force, dynamic range and the
1 Introduction inductive time constant parameters into account of the shock
absorber. The improved performance of the optimized MR
Vehicle suspension systems can be classified into passive, shock absorber was demonstrated using quarter car suspen-
semi-active and active types. These suspension systems sion model under bump and sinusoidal road conditions. Sung
deliver different performance in terms of vehicle handling, et al. [4] studied electronic control suspension (ECS) assem-
bled with cylindrical-type magneto-rheological (MR) shock
absorber. Quarter vehicle system was selected to evaluate
Devdutt (B) · M. L. Aggarwal
Mechanical Engineering Department, YMCAUST, vertical acceleration of car body and tyre deflection in time
Faridabad 121006, India and frequency domain.
e-mail: devdutt.ymca@gmail.com In past, control system based technologies have been
M. L. Aggarwal developed and compared, working on the command signal
e-mail: aggarwalmlal@rediffmail.com for making the practical use of MR shock absorbers in vehi-

123
Devdutt, M. L. Aggarwal

cle suspension system [5,6]. This technology based system In this paper, passenger ride performance of semi-active
is responsible for providing the desired vibration control quarter car system integrated with fuzzy controller is com-
performance through varying the damping capabilities of pared with uncontrolled one. Experimental work is per-
MR shock absorber. Initially, Karnopp et al. studied “Sky- formed with MR shock absorber using various magnitudes
hook” control algorithm for automotive suspension system of input current ranging from 0 A to 1 A. While, polynomial
and the results proved its effectiveness in vibration suppres- model is used for modelling the experimental nonlinear hys-
sion compared to passive system [7]. Other researchers pro- teretic curves of MR shock absorber. This model is helpful
posed Groundhook control for achieving better road hold- in calculating the input current signal to MR shock absorber,
ing ability and Hybrid control which is a comprise between using the desired damping force values and shock absorber
Skyhook and Groundhook control strategy [8]. Based on the piston velocity. A quarter car model with 3 degrees of free-
latest developments in the control system field, many new dom is considered to study the passenger ride comfort and
control methods have been applied in vehicle vibration con- safety issues. Forward as well as Inverse Fuzzy Logic con-
trol and attenuation such as Neuro-fuzzy control [9], Human troller are designed for proper working of physical MR shock
simulated intelligent control [10], Fuzzy Hybrid control [11], absorber in semi-active suspension system. Finally, computer
H∞ control [12] and Particle Swarm Optimization control simulation work is done to evaluate the performance of the
[13], etc. considered systems under bump type of road excitation.
For utilizing the attractive features of MR shock absorbers
in semi-active suspension system related technology, the
prime requirement is the development of forward as well 2 Modeling of test results
as inverse controller. First, the forward controller determines
the desired damping force using the actual damping force For experimental work, the considered MR shock absorber is
values of the physical MR shock absorber. Then the inverse RD-1005-3, manufactured by Lord Corporation, USA. The
controller supply the proper input current to physical MR length of the MR shock absorber in fully compressed state is
shock absorber for producing the actual amount of damping 155 mm. During the extended position, the length is 208 mm.
force matching with the desired damping force. The stroke length of the device is ± 25 mm with the input
A lot of research work has been done during the past voltage requirement of 12 V DC. The force–displacement
few decades for proper modelling the forward dynamic and force–velocity characteristics of the tested MR shock
behaviour of MR shock absorbers. The developed tech- absorber are measured in the laboratory over a range of sinu-
niques for modelling of the forward dynamic behaviour of soidal displacement excitations. The supplied current to MR
MR shock absorbers are generally known as parametric and shock absorber varied from 0 A to 1 A with the increment step
non-parametric models. Parametric models are formed by of 0–0.25–0.5–1 A. For brevity, one particular test results of
the combination of spring and dash-pot type mechanical MR shock absorber at 2.5 Hz and 5 mm excitation under four
elements while non-parametric models employ mathemat- different supplied current values are shown in Fig. 1a,b.
ical expressions using the input-output data of the tested Polynomial model developed by Choi et al. [17] is selected
MR shock absorber. For modelling the experimental data for modelling the nonlinear hysteretic test results of MR
of MR shock absorber, following parametric models have shock absorber. In this model, the hysteresis loop of the
been developed and verified, which includes: Phenomeno- tested MR shock absorber is divided into two curves. The
logical model [14], Visoelastic-plastic model [15] and LuGre upper curve represents negative acceleration state while the
model [16], etc. While the non-parametric models cover: lower curve represents positive acceleration state. The lower
Polynomial model [17], Black-box model [18], Neural net- as well as upper curves can be traced using the power of the
work model [19], Evolving radial basis function networks moving piston velocity v and can be expressed in terms of
[20] and Recursive lazy learning [21], etc. MR shock absorber generated damping force FM R as:
Among the above described models, Phenomenological
model and Viscoelastic-plastic model can very well capture 
n
FM R = ai v i , (1)
the nonlinear behaviour of MR shock absorber. But it is dif-
i=0
ficult to obtain the inverse model from these models. Neural
network model can produce the inverse model effectively where ai is called experimental coefficient, whose value can
but the development of network structure and training the be obtained from the polynomial curve fitting model. In
essential data generate hindrance in its application. In present present case, the order of the fitted polynomial model n = 11
study, polynomial model is selected out of the various avail- is selected based on the trial and error method. The relation-
able models. The selected polynomial model can be used for ship between the coefficient ai i.e. a0 and a4 (the plot of two
controller development due to its ability to generate forward coefficients is shown here for brevity) and supplied current
and inverse dynamics of MR shock absorber. I to the MR shock absorber is plotted in Fig. 2.

123
Fuzzy control of passenger ride performance using MR shock

(a) Based on the plotted Fig. 2, a linear relationship can be


1500 1A established between the coefficients ai and input current I as
follows:
1000 0.5 A
ai = bi + ci I, i = 0, 1, . . . ., 11 (2)
Damping Force (N)

0.25 A
500
0A The calculated values of the coefficients bi and ci are
0 shown in Table 1. The relationship between MR shock
absorber generated damping force and supplied current to
-500 the assembled MR shock absorber can be represented by:
-1000 
11
FM R = (bi + ci I )v i (3)
-1500 i=0
-6 -4 -2 0 2 4 6 The matching characteristics in terms of the force–
Displacement (mm) displacement and force–velocity curves for the selected poly-
(b) nomial model and experimental results are presented in
1500 Fig. 3. It can be observed that the experimental and fitted
1A model curves show very good agreement with each other for
1000
changing values of input current showing the validity of the
0.5 A
Damping Force (N)

500
selected polynomial model.
0.25 A Once the polynomial model is finalized, desired damp-
0 ing force can be obtained by supplying proper input cur-
0A
rent to MR shock absorber by considering the moving piston
-500 velocity. The magnitude of the supplied current to MR shock
absorber can be determined from Eq. 4 as:
-1000
11
Fd − i=0 bi v i
-1500 I = 11 (4)
i=0 ci v
i
-10 -5 0 5 10
Velocity (cm / s)
where I is the supplied current to MR shock absorber, v is
the piston velocity and Fd is the desired damping force signal
Fig. 1 Experimental results (2.5 Hz, ± 5 mm): a force versus displace- generated by designed controller.
ment; b force versus velocity

3 Semi-active quarter car model

Negative acceleration (a0) Negative acceleration (a4)


The quarter car model selected for simulation work is shown
Coefficients of polynomial
Coefficients of polynomial

1000 3
in Fig. 4. Assuming that the assembled tyre of the travelling
750
2 quarter car remains in contact with the road surface for all
500 the time, the dynamic equation of motion of the system can
1 be represented by mathematical equations as follows:
250

0 0 m 1 x¨1 + k2 (x1 − x2 ) + c2 (x˙1 − x˙2 )


0 0.25 0.5 0.75 1 0 0.25 0.5 0.75 1
+FM R = 0 (5)
Positive acceleration (a0) Positive acceleration (a4)
m 2 x¨2 − k2 (x1 − x2 ) − c2 (x˙1 − x˙2 ) + k1 (x2 − x3 )
Coefficients of polynomial

Coefficients of polynomial

0 6

-200 4
+c1 (x˙2 − x˙3 ) − FM R = 0 (6)

-400 2
m 3 x¨3 − k1 (x2 − x3 ) − c1 (x˙2 − x˙3 )
+kt (x3 − xr ) = 0 (7)
-600 0

-800 -2 where m 1 is the passenger seat mass, m 2 is the sprung mass,


0 0.25 0.5 0.75 1 0 0.25 0.5 0.75 1 m 3 is the unsprung mass, k1 and c1 are primary suspension
Input Current (A) Input Current (A)
spring stiffness and shock absorber damping coefficient, k2
Fig. 2 The relationship between ai and supplied current (asterisks and c2 are secondary suspension spring stiffness and shock
Coefficients, line Fitted) absorber damping coefficient, kt is tyre stiffness whereas

123
Devdutt, M. L. Aggarwal

Table 1 Calculated coefficients


(bi and ci ) of fitted curve with Positive acceleration Negative acceleration
values
Coefficients Coefficients Coefficients Coefficients

b0 −87.46 c0 −512.91 b0 58.64 c0 757.99


b1 46.62 c1 144.27 b1 53.71 c1 274.79
b2 8.83 c2 −11.92 b2 −7.78 c2 −51.43
b3 −1.79 c3 0.58 b3 −3.67 c3 −20.67
b4 −0.33 c4 5.53 b4 0.29 c4 2.33
b5 0.08 c5 0.51 b5 0.24 c5 1.59
b6 3.37E-03 c6 −0.26 b6 −3.71E-03 c6 −0.08
b7 −2.64E-03 c7 −0.03 b7 −8.24E-03 c7 −0.05
b8 2.69E-05 c8 4.42E-03 b8 7.53E-06 c8 1.59E-03
b9 4.08E-05 c9 5.55E-04 b9 1.28E-04 c9 7.20E-04
b10 −4.26E-07 c10 −2.63E-05 b10 9.49E-08 c10 −1.06E-05
b11 −2.32E-07 c11 −3.44E-06 b11 −7.31E-07 c11 −3.60E-06

(a)
1500 1A

1000 0.5 A
Damping Force (N)

0.25 A
500
0A
0

-500

-1000

-1500

-6 -4 -2 0 2 4 6
Displacement (mm)

(b)
1500
1A Fig. 4 Semi-active quarter-car model
1000
0.5 A
Damping Force (N)

500 x1 , x2 , x3 and xr denotes the displacement of passenger seat,


0.25 A sprung mass, unsprung mass and input road profile respec-
0
0A tively while FM R is the damping force generated by MR
shock absorber for semi-active control purpose.
-500
Taking dynamic relationship into account, the following
-1000 state variables can be defined

-1500
z 1 = x 1 − x 2 , z 2 = x 2 − x 3 , z 3 = x 3 − xr ,
z 4 = x˙1 , z 5 = x˙2 , z 6 = x˙3 (8)
-10 -5 0 5 10
Velocity (cm / s) where z 1 is secondary suspension deflection, z 2 is primary
Fig. 3 Comparison of model fitted curves and experimental results
suspension deflection, z 3 is the tyre deflection, z 4 is the pas-
(2.5 Hz, ± 5 mm): a force versus displacement; b force versus velocity. senger seat vertical velocity, z 5 is the sprung mass velocity
(lines Experimental, dotted lines Fitted) and z 6 is the unsprung mass velocity while in matrix form

123
Fuzzy control of passenger ride performance using MR shock

Fig. 5 MFs for FFLC a Input (a) (b)


side, V p and Vr el b Output side,

Degree of membership
NB NS ZE PS PB NB NM NS ZE PS PM PB
Fd 1 1

0.5 0.5

0 0
-1 -0.5 0 0.5 1 -3 -2 -1 0 1 2 3

Fig. 6 MFs for IFLC a Input (a)


side, Vr el and Fd b Output side,
(b)
Degree of membership
I NB NS ZE PS PB ZE PS PM PB PVB
1 1

0.5 0.5

0 0
-1 -0.5 0 0.5 1 0 0.5 1

z = [z 1 z 2 z 3 z 3 z 4 z 5 z 6 ]T , u = FM R and w = x˙r (9) to conventional control approach. Therefore, it fits well for
the systems where complete modelling is tedious or ill-
where w is the road input disturbance.
defined such as complex multi-dimensional as well as nonlin-
In state-space form the equations can be represented as:
ear systems and systems with uncertain parameter variations.
In this study, the installed fuzzy logic controller in semi-
ż = Az + Bu + C W W (10)
active quarter car system is the combination of Forward
Fuzzy Logic Controller (FFLC) and Inverse Fuzzy Logic
where
⎡ ⎤ Controller (IFLC). Forward Fuzzy Logic Controller (FFLC)
0 0 0 1 −1 0 uses V p (Passenger seat velocity) and Vr el (Secondary sus-
⎢ 0 0 0 0 1 −1 ⎥ pension velocity) as input variables while Fd (Desired damp-
⎢ ⎥
⎢ 0 0 0 0 0 1 ⎥
⎢ ⎥ ing force) as the output variable whereas Inverse Fuzzy Logic
A = ⎢ − k1 − c1 c1 ⎥,
⎢ m1 0 0 m1 m1 0 ⎥ Controller (IFLC) uses Fd and Vr el as input variables and
⎢ k1 c1 +c2 ⎥
⎣ m 2 − mk22 0 c1
m2 − m2
c2
m2 ⎦ I (Supplied current) as the output variable. The member-
0 k2 kt
− m3 0 c2
− mc23 ship functions (MFs) with selected shapes for Forward Fuzzy
m3 m3

T Logic Controller is shown in Fig. 5 while for Inverse Fuzzy
B = 0 0 0 − m11 m12 0 and Logic Controller is shown in Fig. 6. The applied linguistic
T variables for input and output side of the Forward and Inverse
C W = 0 0 −1 0 0 0 Fuzzy Controller are named using 7 grades as PVB (Positive
represents constant matrices. Very Big), PB (Positive Big), PM (Positive Medium), PS
(Positive Small), ZE (Zero), NS (Negative Small) and NB
(Negative Big) respectively. The rule table for computing Fd
4 Fuzzy logic controller design and I is shown in Table 2.

Fuzzy logic represents a branch of Artificial Intelligence,


which is helpful in solving mathematically complicated and 5 Simulation results and discussion
imprecise real life problems. Fuzzy logic incorporates human
knowledge and decision making abilities into machines. The Simulation work was performed by putting the following
field of fuzzy logic came into existence in 1965, due to the parameter values in the simulink model of the quarter car
efforts and vision of its originator Lotfi Zadeh [22]. Fuzzy system : m 1 = 70 kg, m 2 = 300 kg, m 3 = 40 kg, k1 =
logic controller works on the linguistic variables or human 25, 000 N/m, c1 = 1, 550 N/m/s, k2 = 8, 000 N/m, c2 =
like decision making process. It does not require an exact 800 N/m/s and kt = 180, 000 N/m respectively. The vehicle
mathematical modelling of the process dynamics compared travels with the speed of 40 km/hr and the simulation time

123
Devdutt, M. L. Aggarwal

Table 2 (a) FFLC rule base for computing Fd (b) IFLC rule base for (a)
computing I 0.1

Road Input (m)


(a)
0.05
Vr el /V p NB NS ZE PS PB
0
NB PB PM PS ZE ZE 0 0.5 1 1.5 2 2.5 3 3.5 4
NS PM PS ZE ZE ZE
(b)
ZE PM PS ZE NS NM

Acceleration (m / s )
2
PS ZE ZE ZE NS NM Uncontrolled
5
PB ZE ZE NS NM NB Controlled
0
(b)
-5
Vr el /Fd NB NS ZE PS PB
0.6 0.8 1 1.2 1.4 1.6 1.8
NB PVB PB PB PM ZE
NS PB PM PS ZE PM (c)

Displacement (m)
ZE PB PS ZE PS PB Uncontrolled
0.05
PS PM ZE PS PM PB Controlled

PB ZE PM PB PB PVB 0

-0.05

0.5 1 1.5 2 2.5 3


is fixed as 4 s. The type of road disturbance in the shape of
bump profile is shown in Fig. 7a. (d)
Susp. Deflection (m)

The simulation response of the moving quarter car system 0.05


Uncontrolled
due to road disturbance is shown in Fig. 7b–d. It can be seen Controlled
from Fig. 7b–d that the passenger seat vertical acceleration 0
and displacement response as well as secondary suspension
deflection is considerably improved for controlled system. -0.05
Simulation results show that the controlled system provides 0.5 1 1.5 2 2.5 3

much better results for passenger ride comfort and safety


(e)
compared to uncontrolled one. The desired damping force
Damping Force (N)

200
signal generated by MR shock and the input current signal
supplied to MR shock absorber are shown in Fig. 7 e–f. It can 0
be seen from Fig. 7f that the magnitude of the supplied current
always remains below the maximum value of 1 A, resulting -200
into the lower power consumption by the assembled inverse
0.5 1 1.5 2 2.5 3 3.5 4
fuzzy controller.
The simulation results for acceleration and displacement
(f)
magnitudes of passenger seat as well as secondary suspen-
Input Current (A)

0.2
sion deflection values in terms of RMS (Root Mean Square),
0.15
Max. and Settling time values are presented in Table 3. It can
be observed from Table 3 that the controlled system assem- 0.1
bled with MR shock absorber provides better performance
0.05
in terms of controlling the passenger seat acceleration, dis- 0.5 1 1.5 2 2.5 3 3.5 4
placement and secondary suspension deflection compared to Time (sec)
uncontrolled system. The controlled suspension system also
shows improved performance in terms of settling time values Fig. 7 Response plots of quarter car system: a Bump road excitation
for considered parameters in which controlled system settles b Acceleration of passenger seat c Displacement of passenger seat d
Secondary suspension deflection e Desired damping force signal f Input
down quickly compared to uncontrolled one leading to better current signal
results in terms of passenger ride comfort, safety and vehicle
handling issues.

123
Fuzzy control of passenger ride performance using MR shock

Table 3 RMS, peak and settling


Parameter RMS values Peak values Settling time (sec)
time values for quarter car
model at 40 km/hr Uncontrolled Controlled Uncontrolled Controlled Uncontrolled Controlled

ẍ 1.661 1.452 5.888 5.544 2.990 2.866


x1 0.023 0.019 0.086 0.075 3.077 2.988
x1 − x2 0.010 0.006 0.035 0.017 3.493 2.731

6 Conclusions 8. Ahmadian M, Vahdati N (2006) Transient dynamics of semiac-


tive suspensions with hybrid control. J Intell Mater Syst Struct
17(2):145–153
The application of MR shock absorber in secondary suspen- 9. Eslaminasab N, Biglarbegian M, Melek WW, Golnaraghi MF
sion system of quarter car model is studied in present paper. (2007) A neural network based fuzzy control approach to improve
Polynomial model is formulated using the experimental data ride comfort and road handling of heavy vehicles using semi-active
of MR shock absorber prototype for control system design shock absorbers. Int J Heavy Veh Syst 14(2):135–157
10. Yu M, Dong XM, Choi SB, Liao CR (2009) Human simulated
purpose. Forward Fuzzy controller is designed for provid-
intelligent control of vehicle suspension system with MR dampers.
ing desired damping force signal while Inverse Fuzzy con- J Sound Vib 319(3–5):753–767
troller supplies input current to MR shock absorber in suspen- 11. Rashid MM, Rahim NA, Hussain MA, Rahman MA (2011) Analy-
sion system. The performance of uncontrolled and controlled sis and experimental study of magnetorheological-based damper
for semiactive suspension system using fuzzy hybrids. IEEE Trans
quarter car systems is evaluated by simulation work. The sim-
Ind Appl 47(2):1051–1059
ulation results show the successful application of intelligent 12. Zong LH, Gong XL, Xuan SH, Guo CY (2013) Semi-active H∞
controller in combination with MR shock absorber in vehi- control of high speed railway vehicle suspension with magnetorhe-
cle system. Controlled semi-active suspension system pro- ological dampers. Veh Syst Dyn 51(5):600–626
13. Qazi AJ, Silva CWD, Khan A, Khan MT (2014) Performance
vides better ride comfort to travelling passengers by reducing analysis of a semiactive suspension system with particle swarm
the effect of transmitted vibrations compared to uncontrolled optimization and fuzzy logic control. Sci World J 2014:174102
vehicle system in time domain analysis. 14. Spencer BF Jr, Dyke SJ, Sain MK, Carlson JD (1997) Phenomeno-
logical model for magnetorheological dampers. J Eng Mech ASCE
123(3):230–238
15. Wereley NM, Pang L, Kamath GM (1998) Idealized hysteresis
References modeling of electrorheological and magnetorheological dampers.
J Intell Mater Syst Struct 9(8):642–649
1. Lou Z, Ervin RD, Filisko FE (1994) A preliminary parametric 16. Jimenez R, Icaza LA (2005) LuGre friction model for a magne-
study of electrorheological dampers. Trans ASME J Fluids Eng torheological damper. Struct Control Health Monit 12:91–116
116(3):570–576 17. Choi SB, Lee SK, Park YP (2001) A hysteresis model for the field-
2. Sassi S, Cherif K, Mezghani L, Thomas M, Kotrane A (2005) An dependent damping force of a magnetorheological damper. J Sound
innovative magnetorheological damper for automotive suspension: Vib 245:375–383
from design to experimental characterization. Smart Mater Struct 18. Jin G, Sain MK, Spencer BE Jr (2005) Nonlinear blackbox model-
14:811–822 ing of MR-dampers for civil structural control. IEEE Trans Control
3. Nguyen QH, Choi SB (2009) Optimal design of MR shock Syst Technol 13:345–355
absorber and application to vehicle suspension. Smart Mater Struct 19. Wang DH, Liao WH (2005) Modeling and control of magnetorhe-
18(3):035012 ological fluid dampers using neural networks. Smart Mater Struct
4. Sung KG, Choi SB (2011) Design and control of a MR shock 14(1):111–126
absorber for electronic control suspension. J Korean Soc Precis 20. Du H, Lamb J, Zhang N (2006) Modelling of a magneto-rheological
Eng 28:31–39 damper by evolving radial basis function networks. Eng Appl Artif
5. Dong XM, Yu M, Liao CR, Chen WM (2010) Comparative research Intell 19:869–881
on semi-active control strategies for magneto-rheological suspen- 21. Boada MJL, Calvo JA, Boada BL, Díaz V (2011) Modeling of a
sion. Nonlinear Dyn 59:433–453 magnetorheological damper by recursive lazy learning. Int J Non-
6. Zareh SH, Abbasi M, Mahdavi H, Osgouie KG (2012) Semi-active Linear Mech 46:479–485
vibration control of an eleven degrees of freedom suspension sys- 22. Zadeh LA (1965) Fuzzy sets. Inf Control 8:338–353
tem using neuro inverse model of magnetorheological dampers. J
Mech Sci Technol 26(8):2459–2467
7. Karnopp DC, Crosby MJ, Harwood RA (1974) Vibration control
using semi-active force generators. ASME J Eng Ind 96(2):619–
626

123

You might also like