Journal of Electrical Engineering ISSN 1582-4594 - Vol 17 No 2 (2017)

OPTIMUM PID CONTROLLER DESIGN USING PSO FOR VEHICLE

ACTIVE SUSPENSION SYSTEM CONSIDERING MATLAB SIMULINK
MODELING BASED ROAD PROFILES

Srinivasa Rao GAMPA, Siva Kumar M

Department of Electrical and Electronics Engineering, Gudlavalleru Engineering College, Andhra Pradesh, India
Email: [email protected], Tel: +91 7702427286
D. DAS
Department of Electrical Engineering, Indian Institute of Technology, Kharagpur, India
Email: [email protected], Tel:: +91 3222 283042

Abstract: The vehicle active suspension system is mainly disturbances on the passengers and hence they are
used in the automobile industry to improve ride comfort, more efficient compared to passive suspension
road handling ability under different road disturbance system and semi-active suspension systems [1-14].
conditions. In the present paper particle swarm
optimization (PSO) based Proportional-Integral- Researchers have proposed many techniques
Derivative (P-I-D) controller is proposed considering the based on conventional and artificial intelligence
objective of minimizing the sprung mass acceleration based techniques for vehicle suspension system.
when the vehicle is subjected to various road conditions. Gordon [1] proposed dynamic optimization
Matlab Simulink based quarter vehicle and road bump technique for the semi-active suspension quarter
disturbance models are developed for the dynamic vehicle model to minimize body acceleration and
analysis of the vehicle suspension system. Simulation
suspension deflections. Hayakawa et al. [2] have
results are demonstrated to show the advantage of the
proposed PSO based P-I-D controller in terms of proposed output feed control for automobile
improving the dynamic performance of the active suspension system based on kanes dynamics in order
suspension system regarding body acceleration, velocity, to reduce controller complexity. Yang et al. [3] have
position and providing good tire-road contact compared designed two-degree-of-freedom controller in order
to Ziegler-Nichols tuning method and passive suspension to minimize disturbances caused by the vehicle
system. engine body system. Lu Sun [4] has developed an
Key words: Particle swarm optimization Quarter vehicle optimum design methodology based on stochastic
model, Suspension system, Suspension deflection, Road profiles process theory to minimize the impact of tire load by
the vehicles on the road pavement. Al-Holou et al.
1. Introduction [5] have proposed adaptive fuzzy based controller to
Modern trends in technology have been determine the designer objective such that it
drastically changed the automobile industries. provides robust and adaptive control of desired
Suspension system plays a critical role in improving parameters of vehicle suspension system. Baumal et
ride comfort and road-handling ability. The main al. [6] proposed genetic algorithm based
function of suspension system is to minimize the optimization technique for minimizing the
vertical acceleration of the vehicle body that is passenengers seat acceleration for active suspension
transmitted to the occupants in order to achieve ride system control under road disturbance conditions.
comfort and also keep the tires in contact with road Kuo and Hseng [7] have developed fuzzy PI/PD
during disturbance conditions of the vehicle. Vehicle controller based on genetic algorithm in order to
suspension system can be controlled by using three provide optimal decision-making rules for different
basic approaches which are classified as passive, road conditions. Al-Holou et al.[8] have developed
semi-active and active suspension systems. Active a robust intelligent nonlinear controller based on the
suspension system employs actuators to provide combination of sliding mode, neural network and
desired control action between the vehicle sprung fuzzy logic techniques to handle complex
body and wheel axle to minimize the impact of road uncertainties in active suspension system. Rattasiri

286
Journal of Electrical Engineering ISSN 1582-4594 - Vol 17 No 2 (2017)

and Halgamuge [9] have presented a new type of In this paper particle swarm optimization based
hierarchical fuzzy classifying control system in optimum Proportional-Integral-Derivative controller
order to avoid repetitive defuzzification process in is proposed for minimizing vertical body
conventional fuzzy systems and to minimize the acceleration and to provide better road handling
computational time for active suspension system ability for vehicle active suspension system
control. Precup et al. [10] proposed combination of considering MATLAB simulink modeling based
Takagi Sugeno and interpolative fuzzy controllers road disturbances.
for tire slip control in anti lock braking systems.
2. State Space Modeling of Active Suspension
Tung et al.[11] have proposed a new exponential System for Quarter Vehicle Model
stabilization methodology for active suspension
system control based on particle swarm optimization Suspension System can be modeled as two degree
(PSO) algorithm for twin-shaft vehicle model with freedom system, approximating as a dual spring-
three-degree-of-freedom with bounded uncertainties. mass-damper with controlled actuator to represent
Ghafouri et al. [12] have developed a PI controller the quarter-vehicle active suspension system model
for quarter car model of active suspension system in as shown in Fig. 1. The system consists of a spring,
which the gains are tuned by using PSO to optimize damper and actuators, which provide the desired
body acceleration and tire deflection. Soleymani et control force in the suspension system, connecting
al. [13] have proposed a multi-objective adaptive the, vehicle body to ground via the wheel tire spring.
fuzzy controller and parameters are tuned for
various traffic conditions via a multi-objective
genetic algorithm based optimization approach.
Huerel et al. [14] proposed a PSO based fuzzy PD
controller considering the objective of minimizing
sprung mass acceleration for active suspension
system.
Precup et al. [15] proposed an adaptive
gravitational search algorithm for optimal tuning of
fuzzy controlled servo systems considering the
objective of minimizing process gain variations and
sensitivity. Rmirez et al.[16] proposed a novel
optimization technique based on Bayes theory
considering the objective of minimum error rate
threshold for removing binary artifacts from hand
written and printed documents. Ragab et al. [17]
proposed a fuzzy based multiobjective algorithm for
reactive power management problem considering Fig.1 Quarter vehicle active suspension system
both technical and economical aspects. Hongyi et al.
The dynamics of the sprung mass and unsprung
[18] have designed output-feedback and state-
mass of the quarter-vehicle suspension model can be
feedback based sampled-data controllers based on
lyapunov stability theory via Takagi-Sugeno fuzzy derived using Newton’s second law, and are given
control approach. Khmelev et al.[19] proposed a by the following equations.
hybrid metaheuristic optimization technique for MSMZ&&SM + KS (ZSM − ZUM) + Ca(Z&SM − Z&UM) − Fa = 0 (1)
optimum split delivery vehicle routing combining
neighborhood decent and stochastic tabu search MUMZ&&UM + KS (ZUM −ZSM) +Ca (Z&UM − Z&SM) + Kt (ZUM − Zr ) + Fa =0 (2)
methods.Brezas et al. [20] developed an algorithm
based on Hamilton Jacobi Bellaman for obtaining Where,
simultaneously optimizing ride and road handling MSM: Sprung mass
behaviuor of semi active suspension system. Lin et
MUM: UnSprung mass
al. [21] have proposed adaptive sliding mode
technique for electro magnetic suspension system ZSM: Displacement of sprung mass
with linear switched reluctance actuator to ZUM: Displacement of unsprung mass
compensate suspension system nonlinearities and KS: Spring stiffness constant Suspension
external road disturbances. Cd: Damping coefficient
Kt: Spring stiffness Constant of Tire

287
Journal of Electrical Engineering ISSN 1582-4594 - Vol 17 No 2 (2017)

Fa: Actuator Control force the road disturbances are of bump roads. For
Zr: Displacement of tire dynamic response analysis of the vehicle suspension
system on rough road, the road disturbances are
modeled with simple sinusoidal functions. In the
The active suspension system parameters [14] are present analysis road disturbances with one bump
given in Table-A in the Appendix A. and two bumps are considered for the performance
In the present work a proportional integral derivative analysis of the vehicle suspension system.
controller proposed for improving the vehicle 3.1 Road disturbance with one bump (Rd1)
dynamics.
The road disturbance with one bump can be modeled
The P-I-D control law can be written as follows by the following equation.
d (e(t ))
Fa = K p e(t ) + K i ∫ e(t )dt + K d (3)
dt
A   t  
Where Kp, Ki and Kd are proportional, integral and  1 - cos 2π   , for Tb1 ≤ t ≤ 2Tb1 (8)
derivative gain constants of the P-I-D controller. Rd1(t) =  2    Tb1   
Here the sprung mass acceleration is taken as feed 
0 Otherwise
back and hence the error input for the P-I-D
controller can be defined as follows: In the Eq.(8), A is the height of the bump and Tb1
is the duration of the bump. The magnitude of the
e(t ) = Rref − Z&&SM (4)
duration of the bump is defined as the ratio of bump
The state variables for active suspension system are length (L) to the velocity of the vehicle (V). In the
chosen as follows: present work it is considered that the road bump
disturbance is generated between the time Tb1 and
X 1 = Z SM , X 2 = ZUM , X 3 = Z& SM , X 4 = Z&UM ,
2Tb1 seconds and for the remaining period plain road
X 5 = Z&&SM and X 6 = K i ∫ e(t ) dt profile is assumed.
The state space model can be described by the 3.2 Road disturbance with two bumps (Rd2)
following state equation, The road disturbance with two bumps of different
magnitudes is modeled using the following Eq.(9).
X& = PX + ΓR (5)
A   t  
Where X and R are state and input vectors  1 - cos 2π   , for Tb1 ≤ t ≤ 2Tb1
2    Tb1   
respectively and P and Γ are constant matrices of  (9)
B    t   
Rd 2 (t ) =  1 - cos 2π   , for 8Tb2 ≤ t ≤ 9Tb2
active suspension system with the P-I-D controller.
 2    T  
  b2  
X ' = [ X 1, X 2 , X 3 , X 4 , X 5 , X 6 ] 

(6) 0 Otherwise
R' = [ Rref Rd ] (7) In the Eq.(9) it is considered that the height of the
initial road bump disturbance is A during the first
In the Eq.(7) Rref and Rd are reference input and
disturbance time interval between Tb1 and 2Tb1
road disturbance of the active suspension system. seconds and the second bump disturbance of height
The X ' and R' are transpose matrices of X and R. B is considered during the time interval between
The matrices P and Γ are given in Appendix-B. 8Tb2 and 9Tb2 seconds.
4. Simulink Modeling of Active Suspension
3. Modeling of Road Profiles System for Quarter Vehicle Model
In order to investigate the suspension dynamics In order to perform dynamic performance
with respect to ride comfort and handling capability analysis of the active suspension system subject to
of the vehicle different models of the road profiles
road disturbances in the present work Matlab
are taken into consideration. Generally the road
disturbances which appear relatively for very short simulink models are developed for quarter car model
duration and with high impact will seriously effect and road disturbances from the equations developed
passengers ride comfort and may lose holding in the previous two sections for the active
capability of the road surface. In general many of suspension system.

288
Journal of Electrical Engineering ISSN 1582-4594 - Vol 17 No 2 (2017)

4.2 Simulink model of road disturbance with two

bumps
Matlab Simulink model of road disturbance with
two bumps is developed from Eq.(9) and is shown
is Fig.4.

Fig. 2. Simulink model of quarter vehicle active

suspension system. Fig. 4. Simulink model road disturbance with two bumps.

In the present analysis minimization of vehicle body In the Fig.4 the values of T1 and T2 are considered equal
acceleration using PSO based PID controller is considered to bump starting time Tb1 and bump ending time 2Tb1
and it is incorporated in the simulink model. Matlab during the first road disturbance interval as described in
simulink models are developed for road disturbances for Eq.(9) respectively. The values of T3 and T4 are
analyzing the performance of active suspension system. considered equal to 8Tb2 and 9Tb2 respectively during
In the present work simulink models are developed for the second road disturbance interval with two bumps.
two types of road disturbances considered one with single
5. Optimum PID Gains for Vehicle Active
bump and the other with two bumps as described by
Suspension System Using PSO
Eq.(8) and Eq.(9) in the previous section.
Particle Swarm Optimization (PSO) is a population-
4.1 Simulink model of road disturbance with one based search algorithm proposed by Kennedy and
bump Eberhant and is motivated by simulation of social
Simulink model of road disturbance with single bump is behavior of bird flocking, fish schooling and swarming
developed from Eq.(8) is shown is Fig.3. theory [22]. In PSO each particle is associated with two
set of factors the velocity and position. The velocity
factor is updated based on particles own experience and
social experience within the swarm and position of the
particle is updated using velocity of the particle. The
particles move in the multi-dimensional search space
collectively based on the objective to be optimized to find
the global best in total swarm in the given search space.
The velocity and position of the particles are updated
using the following equations.
k+1 k k
Vi =Wi Vi + c1r1(Pbest ki − Sik ) + c2r2 (Gbestki − Sik ) (10)

S ik +1 = Vi k + S ik (11)
Fig. 3. Simulink model road disturbance with one bump.
In the Fig.3 the values of T1 and T2 are considered equal The values of C1 and C2 are in the Eq.(10) are
to bump starting time Tb1 and bump ending time 2Tb1 as acceleration constants set to a constant value of 2.
described in Eq.(8) respectively. The values of r1 and r2 are randomly generated

289
Journal of Electrical Engineering ISSN 1582-4594 - Vol 17 No 2 (2017)

numbers inbetween 0 and 1. The velocity and

position of ith particle of (k+1)th iteration Vi k +1 and
S ik +1 are updated using Eq.(10) and Eq.(11). The
inertia weight (W) linearly decreases from 0.9 to 0.4
as the algorithm proceeds from one iteration to
another iteration. The velocity is limited to
predetermined minimum and maximum values. The
Pbest ki , Gbest ki are population best and global best of
ith particle and kth iteration of PSO algorithm[22].

Comparing with other evolutionary based

algorithms particle swarm optimization algorithm is
more popular because of it is simple in
implementation and ablility to converge to a good
solution with minimum number of iterations. In the
present work the optimum gains of Proportional-
Integral-Derivative (P-I-D) controller are obtained
using PSO considering the objective of minimization
of vehicle sprung body accelarion which is
described by Eq.(12).
t
J = ∫ ( Z&&s (t)) 2 dt (12)
0

The PSO algorithm proceeds in the direction of

maximization of fitness function and hence the
fitness function JF is considered described by the
following Eq.(13) for obtaining the optimum
solution.
K1 (13)
JF =
1 + K2 J
The constants K1 and K2 are scaling factors and
here both the values are set to a constant value of
100.
The flow diagram for obtaining optimum P-I-D Fig. 5. Flow chart for obtaining optimum PID gains
gains using PSO for active suspension system is using PSO for active suspension system
shown in the following Fig.5.
The step by step algorithm for PSO based Step-4: If the fitness value of the particle from
optimization for vehicle suspension system is given current population is better than the corresponding
below. previous population set the current value as Pbest.

Step-1: Firstly the particles of the swarm are Step-5: Set the best fitness value from the all
initialized by the randomly generated population for previous iterations as Gbest.
the P-I-D controller gains and set the velocity
positions to a constant value. Step-6: Determine velocity and position of each
particle using Eq.(10) and Eq.(11) for next iteration.
Step-2: Set the iteration count to one.
Step-7: Increment the iteration count by one and if
Step-3: Calculate the fitness function values for the iteration count is less than the maximum value
each particle using Eq.(13). Go To Step-3.

290
Journal of Electrical Engineering ISSN 1582-4594 - Vol 17 No 2 (2017)

6. Results and Discussions In the second case the road disturbance with two
bumps of height (A) 0.05m and height (B) 0.075m is
In the present work for improving the dynamic
considered. It is assumed that the length of the first
performance of the vehicle active suspension system
bump is 5m and second bump is 6m, and the vehicle
PSO based P-I-D controller is proposed. The
is moving with a velocity (V) of magnitude 36km/h.
optimum gains are obtained considering the
The duration of first bump Tb1 calculated as the
objective of minimizing vehicle body acceleration is ratio of bump length to velocity is to 0.5 seconds in
considered. Matlab Simulink models are developed
the case of second bump is 0.75 seconds. The
for road disturbances to facilitate the analysis using
vehicle is crossing the first bump inbetween the time
Simulink modeling. In the present analysis two types
interval 0.5 to 1.0 seconds and inbetween the time
of road bumps disturbances are considered one with
interval 6.0 to 6.75 seconds.
single bump and another with two bumps of
different magnitude of height and time duration. The optimum gains of the P-I-D controller are
obtained using PSO considering the objective of
In the first case the road disturbance with single
minimizing vehicle sprung body acceleration for
bump of height (A) 0.05m is considered. It is
single bump disturbance conditions and the P-I-D
assumed that the length of the bump (L) is 5m and
gains are Kp = 4912.38, Ki = 2935.90 and Kd =
the vehicle is moving with a velocity (V) of
22.61.
magnitude 36km/h. The bump duration Tb1 is
calculated as (L/V) and is equals to 0.5 seconds,
and it is considered that vehicle is crossing the bump
inbetween the time interval 0.5 to 1.0 seconds. The
road bump disturbance with single bump is shown in
the following Fig.6.

Fig.8 Sprung body position with single bump disturbance.

Fig. 6. Road disturbance with single bump.

The road disturbance with two bumps is shown in
the following Fig.7.

Fig. 7. Road disturbance with two bumps. Fig.9. Sprung body velocity with single bump
disturbance.

291
Journal of Electrical Engineering ISSN 1582-4594 - Vol 17 No 2 (2017)

Fig.10. Sprung body acceleration with single bump

disturbance.
Fig.12. Sprung body velocity with two bump
The performance of the active suspension system disturbances.
with PSO based P-I-D controller is compared with
conventional Ziegler-Nichols (ZN) tuning method.
The dynamic performance improvement of proposed
PSO based P-I-D controller is shown in Fig.8, Fig.9
and Fig.10 in the case of road disturbance with
single bump.
From Fig.8, Fig.9 and Fig.10 it can be observed
that the position, velocity and acceleration of vehicle
sprung body subjected to single bump disturbance in
the case of active suspension system with PSO based
proposed P-I-D controller is much improved
compared to passive suspension system and active
suspension system with conventional Ziegler-
Nichols method.

Fig.13. Sprung body acceleration with two bump

disturbances.
The P-I-D controller gains obtained with proposed
PSO based methodology for single bump
disturbance are used for analyzing the performance
of the active suspension system in the case of road
disturbance with two bumps to test the sensitivity of
the P-I-D controller gains. From Fig.11, Fig.12 and
Fig.13 it can be observed that the dynamic
performance of the active suspension system is
much better in the case of sprung mass position,
Fig.11. Sprung body position with two bump velocity and acceleration compared to Ziegler
disturbances. Nichols based active suspension system method and
passive suspension system without controller.

292
Journal of Electrical Engineering ISSN 1582-4594 - Vol 17 No 2 (2017)

From the simulation results it can be observed that

the performance of proposed PSO based P-I-D
controller is much better in comparison with the
passive system and active system based ZN-method
in terms of position, velocity and acceleration and
for providing good road-tyre contact during the
bump road disturbance conditions.
Appendix-A
Table A: Parameters of Quarter Vehicle Model [14]

Fig.14. Suspension deflection with one bump Sprung mass Msm = 250Kg
disturbance.

Unsprung mass Mum = 50Kg

Suspension damping Cd=1500Ns/m

Suspension stiffness Ks = 16000N/m

Tire stiffness Kt = 160000N/m

Appendix-B
Fig.15. Suspension deflection with two bump The constant matrices P and Γ corresponding to
disturbances.
active suspension system state space model are
From Fig.14 and Fig.15 it can be observed that given below.
during the bump disturbance intervals the negative
deflection is more in the case of active suspension
system with proposed controller The higher negative
deflection indicates better road holding capacity and
good tyre-road contact during the disturbance
conditions.
7. Conclusions
In this paper PSO based optimum P-I-D controller
is proposed for active suspension system for quarter
vehicle model in order to achieve the objective of
suspension system of better ride comfort and road
handling ability during road disturbance conditions.
In the present analysis two types of road
disturbances, one with single bump and the other
with two bumps are considered and Matlab Simulink
models of single bump and two bump road
disturbances are developed along with quarter
vehicle model for facilitating the dynamic response
analysis in Matlab Simulink environment. The
optimum P-I-D gains are obtained using PSO and
Matlab simulink models developed considering the
objective of minimizing sprung mass acceleration.

293
Journal of Electrical Engineering ISSN 1582-4594 - Vol 17 No 2 (2017)

Electronics, XLVI(1999), December 1999, No. 6,

p.1051-1056.

8. Al-Holou,N., Lahdhiri,T., Sung Joo,D., Weaver.J,

Al-Abbas,F.: Sliding Mode Neural Network
Inference Fuzzy Logic Control for Active
Suspension Systems. In: IEEE Transactions on
Fuzzy Systems, X (2002), No. 2. April 2002, p.234-
246.

9. Rattasiri,W., Halgamuge,S. zK.: Computationally

Advantageous and stable Hierarchical Fuzzy
Systems for Active Suspension. In: IEEE
Transactions on Industrial Electronics. L (2003),
No. 1. February 2003, p.48-61.

10. Precup, R. E., Preitl,S., Balas. M, Balas,V.: Fuzzy

controllers for tire slip control in anti-lock braking
systems. In: Proceedings of IEEE International
Conference on Fuzzy Systems, July 25-29, 2004,
References Budapest, Hungary, Vol.III, p. 1317-1322.
11. Tung,S.L., Juang,Y.T., Lee,W.H., Shieh,W.Y.,
1. Gordon, T.J: Non-Linear Optimal Control of a Semi-
Wu,W.Y.: Optimization of the exponential
Active Vehicle Suspension System. In: Chaos,
stabilization problem in active suspension system
Solitons and Fractals V (1995), No. 9, September
using PSO. In: Expert Systems with Applications,
1995, p. 1603-1617.
XXXVIII(2001),October 2001,p.1444-1451.
2. Hayakawa. K., Matsumoto,K., Yamashita,M.,
12. Ghafouri, J., Sakhavati, A., Ramin, J.,
Suzuki,Y., Fujimori.K., H. Kimura,H.: Robust H-
infinity Output Feedback Control of Decoupled Gharehpetian, G.B.: PI Controller for Quarter
Automobile Active Suspension Systems. In: IEEE Vehicle Active Suspension System with Optimized
Gains Using PSO Algorithm. In: 2nd International
Transactions on Automatic Control, XLIV (1999),
Conference on Control, Instrumentation and
No. 2., February 1999, p.392-396.
Automation, December 27-29, 2011, Shiraz, Iran,
3. Yang,J., Suematsu,Y., Kang,Z.: Two-Degree-of- p. 379-383.
Freedom Controller to Reduce the Vibration of
13. Soleymani, M., Montazeri-Gh, M., Amiryan,R.:
Vehicle Engine-Body System. In: IEEE Transactions Adaptive fuzzy controller for vehicle active
on Control Systems Technology, IX (2001), No. 2., suspension system based on traffic conditions. In:
March 2001, p.295-304. Transactions on Mechanical Engineering, Scientia
Iranica B, XIX (2012), No.3, June 2012, p.443-453.
4. Sun,L.: Optimum design of “road-friendly” vehicle
suspension systems subjected to rough pavement 14. Hurel, J., Mandow,A., A. Garcia-Cerezo,A: Tuning
surfaces. In: Applied Mathematical Modeling, a Fuzzy Controller by Particle Swarm Optimization
XXVI (2002), No.5, May 2002, p.635-652. for an Active Suspension system. In: IEEE 38th
Annual Conference on Industrial Electronics
5. Al-Holou, N., Weaver, J., Sung, J.: Adaptive Fuzzy Society, October 25-28, 2012, Montreal, Canada,
Logic Based Controller for an Active Suspension p.2524-2529.
System. In: IEEE Proceedings of the 39th Midwest
symposium on circuits and systems, August 18-21, 15. Precup, R.E., David, R.C., Petriu, E.M, Preitl, S.,
1996, Ames, IOWA,USA, Vol. II, p. 583-586. Radac,M.B: Fuzzy logic-based adaptive
gravitational search algorithm for optimal tuning of
6. Baumal,A.E., McPhee,J.J., Calamai,P.H.:Application fuzzy controlled servo systems. In: IET Control
of genetic algorithms to the design optimization of Theory & Applications, VII (2013), January 2013,
an active vehicle suspension system. In: Computer No. 1, p. 99-107.
Methods in Applied Mechanics and Engineering,
16. Ramírez-Ortegon, M.A., Margner, V., Erik Cuevas,
CLXIII(1998), No. 1, September 1998, p. 87-94.
Raul Rojas: An optimization for binarization
7. Kuo,Y.P., Tzuu-Hseng, Li,S.: GA-Based Fuzzy methods by removing binary artifacts. In: Pattern
PI/PD Controller for Automotive Active Suspension Recognition Letters, XXXIV (2013), August 2013,
System. In: IEEE Transactions on Industrial No. 11, p. 1299-1306.

294
Journal of Electrical Engineering ISSN 1582-4594 - Vol 17 No 2 (2017)

17. Ragab El Sehiemy, Adel Abou El-Ela, AbdulAllah

Shaheen: Multi-objective fuzzy-based procedure for
enhancing reactive power management.In: IET
Generation, Transmission & Distribution,
VII(2013), No. 12, December 2013, p. 1453-1460.
18. Hongyi,L., Jing,X., Lam H.K., Shi.P.: Fuzzy
Sampled-Data Control for Uncertain Vehicle
Suspension Systems. In: IEEE Transactions on
Cybernetics, XLIV(2014), No. 7. July 2014, p.1111-
1126.
19. Khmelev,A., Kochetov,Y.: A hybrid local search for
the split delivery vehicle routing problem. In:
International Journal of Artificial Intelligence,
XIII(2015), No. 1, March 2015, p. 147-164.
20. Brezas, P., Malcolm, C. S., Hoult, W.: A clipped-
optimal control algorithm for semi-active vehicle
suspensions: Theory and experimental evaluation.
In: Automatica, LIII (2015), March 2015, p.188-
194.
21. Lin,J., Cheng,K.W.E., Zhang,Z., Cheung N.NC.,
Xue.,X: Adaptive sliding mode technique-based
electromagnetic suspension system with linear
switched reluctance actuator. In: IET, Electrical
power applications,IX (2015), No.1, January 2015,
p.50-59.
22. Kennedy,J., Eberhart,R., Particle Swarm
Optimization. In: IEEE International Conference on
Neural Networks, 27 November - 1st December,
1995, Perth, WA, Vol. IV , p. 1942-1948.

295