ISA Transactions 67 (2017) 222–232

Contents lists available at ScienceDirect

ISA Transactions
journal homepage: www.elsevier.com/locate/isatrans

Research article

Fractional order PID control design for semi-active control of smart base-
isolated structures: A multi-objective cuckoo search approach
Abbas-Ali Zamani a, Saeed Tavakoli a,n, Sadegh Etedali b
a
Department of Electrical Engineering, University of Sistan and Baluchestan, Zahedan, Iran
b
Department of Civil Engineering, Birjand University of Technology, Birjand, Iran

art ic l e i nf o a b s t r a c t

Article history: Fractional order PID (FOPID) controllers are introduced as a general form of classical PID controllers using
Received 8 December 2015 fractional calculus. As this controller provides good disturbance rejection and is robust against plant
Received in revised form uncertainties it is appropriate for the vibration mitigation in structures. In this paper, an FOPID controller
18 April 2016
is designed to adjust the contact force of piezoelectric friction dampers for semi-active control of base-
Accepted 5 January 2017
isolated structures during far-field and near-field earthquake excitations. A multi-objective cuckoo search
Available online 20 January 2017
algorithm is employed to tune the controller parameters. Considering the resulting Pareto optimal front,
Keywords: the best input for the FOPID controller is selected. For seven pairs of earthquakes and nine performance
Base-isolated structure indices, the performance of the proposed controller is compared with those provided by several well-
Semi-active control
known control techniques. According to the simulation results, the proposed controller performs better
Piezoelectric friction damper
than other controllers in terms of simultaneous reduction of the maximum base displacement and story
Fractional order PID controller
Multi-objective cuckoo search acceleration for various types of earthquakes. Also, it provides acceptable responses in terms of inter-
story drifts, root mean square of base displacements and floor acceleration. In addition, the evaluation of
robustness for a stiffness uncertainty of ±10% indicates that the proposed controller gives a robust
performance against such modeling errors.
& 2017 ISA. Published by Elsevier Ltd. All rights reserved.

1. Introduction energy for strong earthquakes. Therefore, a controllable sliding

force is required to adjust the essential amount of energy dis-
One of the practical techniques for simultaneous reduction of sipation for different levels of earthquakes. A variable friction
drift and acceleration of stories during earthquake excitation is damper is able to adjust the sliding force by an actuator. Several
seismic isolation [1]. However, a base-isolated structure may experimental and theoretical studies have been conducted on
experience an excessive response during a near-field earthquake these dampers in recent years [4–8].
with a long-period pulse [2]. In order to improve the perfor- Piezoelectric friction dampers, as semi-active control devices,
mance and safety of the base-isolated structures during near- are variable friction dampers. They can be used to suppress
field earthquake excitations, the use of passive, semi-active and structural vibrations [9]. They operate with low energy and do not
active devices has been proposed [3]. The large deformations of cause instability of the controlled structures [10]. To enhance the
the isolation bearings during strong earthquakes can be reduced performance of seismic isolation systems during near-field earth-
using passive friction dampers, as energy dissipation systems. quakes, Lu and Lin [11] proposed a piezoelectric smart isolation
system, a semi-active isolation system equipped with PFD. How-
However, they can increase both internal deformations and ab-
ever, an effective control algorithm is needed to take advantage of
solute accelerations of the superstructure floors. This issue limits
the full capabilities of PFDs in smart base-isolated buildings. Lu
many advantages of the isolators [2,3]. Moreover, passive devices
et al. [12] studied a PSIS controlled by a fuzzy controller. Ozbulut
such as passive friction dampers cannot adjust the slip force in
and Hurlebaus [13] suggested two fuzzy logic controllers for a
real time. When the sliding force is adopted too large, dampers
base-isolated structure equipped with PFDs. To modulate the
will not be able to dissipate the energy for weak or moderate
control force of PFDs, Ozbulut et al. [14] proposed an adaptive
earthquakes and, thus, they will not slide. On the other hand, for
control strategy as well as a direct fuzzy controller with self-tuning
a small sliding force, dampers will not be able to waste enough scaling factors based on neural networks. A hierarchic fuzzy con-
trol algorithm, in which a supervisory fuzzy controller governs a
n
Corresponding author. sublevel fuzzy controller by altering its input normalization factors
E-mail address: [email protected] (S. Tavakoli). according to current level of ground motion, was proposed in [15].

http://dx.doi.org/10.1016/j.isatra.2017.01.012
0019-0578/& 2017 ISA. Published by Elsevier Ltd. All rights reserved.
 A.-A. Zamani et al. / ISA Transactions 67 (2017) 222–232 223

Nomenclature MR magneto-rheological
CS cuckoo search
P proportional MOCS multi-objective cuckoo search
PD proportional-derivative FP fault-parallel
PI proportional-integral FN fault-normal
PID proportional-integral-derivative SVFLC supervisory fuzzy logic controller
FOPID fractional order proportional-integral-derivative SOFLC self-organizing fuzzy logic controller
VFD variable friction damper GHFLC genetic algorithm hierarchic fuzzy logic control
PFD piezoelectric friction damper RMS root mean square
PSIS piezoelectric smart isolation system

To simultaneously reduce the base displacement and super- better performance in comparison with standard PID controllers.
structure responses of the base-isolated structure during seismic An FOPID was designed for active vibration control of a strip in a
excitations, the GA was employed to optimize the supervisory continuous galvanizing line by Koo et al. [35]. FOPID parameters
fuzzy controller and the preload of PFDs. were determined using particle swarm optimization and its per-
Due to their remarkable effectiveness and simplicity of im- formance was compared with that of the standard PID controller.
plementation, PID controllers are widely used in various en- The problem of semi-active control of smart base-isolated
gineering applications. However, there are few studies on PID structures during far-field and near-field earthquake excitations
control design for seismic control of structures. Aguirre et al. [16] includes many conflicting design objectives. Such optimization
designed a PI controller to minimize the structural responses of a problems have many local optimums and require a heavy small-
3-story building equipped with Magneto-Rheological dampers. To scale search. To cope with this, a meta-heuristic multi-objective
reduce superstructure responses and drifts of a base-isolated optimization algorithm keeping enough diversity of the popula-
building under near-field earthquake excitations, a control scheme tion is required. Also, the accuracy of solutions is an important
was proposed by Subasri et al. [17]. It was based on a combination issue. The CS is able to explore the search space more accurately
of discrete PID controller and discrete direct adaptive neural and can perform better for challenging objective functions. For
controller. Nigdeli [18] studied the effect of feedback control base-isolated buildings, the main goal of seismic control is to re-
strategies on active structures using PID controllers under earth- duce the isolation displacement without any increase in the su-
quake excitations. Yu et al. [19] developed PD/PID controllers for perstructure acceleration. As a result, a trade-off between these
active vibration control of a structure equipped with an active conflicting objectives should be made.
mass damper. Etedali et al. [3] employed optimal PD/PID con- Due to successful application of FOPID controllers for the vi-
trollers to adjust the contact force of PFDs in a 5-story isolated bration mitigation, it may handle the problem of seismic vibration
structure. In addition, Etedali et al. [20] developed an independent mitigation too. However, no research work, to the best of knowl-
robust modal PID control approach for seismic control of edge of authors, has been carried out on the seismic control of
structures. base-isolated structures using FOPID controllers. In this paper, an
Vibration suppression is a major problem in various domains. FOPID controller using multi-objective cuckoo search algorithm is
In order to overcome this problem, a number of techniques have designed to adjust the contact force of PFDs considering the semi-
been used, however, only a few of them have addressed this issue active control of the building during far-field and near-field
from the fractional calculus perspective. The main advantage of a earthquake excitations.
fractional order controller is its ability to meet more design spe- The main contributions of this paper are as follows. Consider-
cifications at the same time. Wang and Zheng [21] showed that ing the semi-active control of the building during far-field and
fractional order integrators and differentiators could improve the near-field earthquake excitations, an FOPID controller is designed
stability of vibration systems. A fractional order derivative con- to adjust the contact force of PFDs. Using a challenging base-iso-
troller for seismic mitigation of structures equipped with viscoe- lated structure equipped with PFDs, the performance of the de-
lastic dampers was proposed by Muresan et al. [22]. signed FOPID controller is evaluated. Considering the main goal of
Through combining the idea of PID control with the theory of seismic control, reducing the isolation displacement without any
fractional calculus, a new controller named fractional order PID increase in the superstructure acceleration, a trade-off between
was introduced. During the past two decades, a considerable at- conflicting structural responses is made using the MOCS. The im-
tention has been paid to the design of FOPID controllers using portant feedbacks in seismic control of base-isolated buildings are
analytical [23–26] and heuristic [27–32] methods. An FOPID con- the isolation displacement and roof acceleration. Considering
troller performs better than a standard PID controller in terms of three candidate feedbacks, i.e. the isolation displacement, isolation
design flexibility, robustness, disturbance rejection, and steady- velocity, and roof acceleration, the best input for FOPID controller
state error elimination. Therefore, it can be a suitable choice for is proposed according to the resulting Pareto optimal front. Con-
the vibration mitigation and control. Recently, few research works sidering 14 real-data earthquakes and nine performance indices,
focusing on the application of FOPID for the vibration suppression the performance of the proposed controller is compared with
problem have been published [33–35]. To improve the delay- those given by other control strategies, namely the maximum
margin of the non-collocated vibration control of piezo-actuated passive operation of PFDs, modified clipped-optimal controller
flexible beams, an FOPID was proposed by Sangpet et al. [33]. [14], supervisory fuzzy logic controller [13], GA-based self-orga-
Experimental results showed that the fractional order controller nizing fuzzy logic controller [13], genetic algorithm hierarchic
could provide a better stability robustness than the integer order fuzzy logic control [15] and optimal PID controller [3]. It is shown
counterpart. Using a multi-objective genetic algorithm, Gad et al. that, opposite to other control strategies, the proposed FOPID
[34] proposed an FOPID to determine the desired damping force controller is able to reduce the isolation displacement without any
for semi-active MR damped seat suspension in mechanical appli- increase in the superstructure acceleration for all given earth-
cations. It was shown that the proposed controller could offer a quakes. Considering all given earthquakes, the robustness of the
224 A.-A. Zamani et al. / ISA Transactions 67 (2017) 222–232

proposed controller for a stiffness uncertainty is evaluated. N (t ) = Npre + CPZ V (t ) (5)

The rest of the paper is organized as follows. The model of a
base-isolated structure equipped with PFDs is introduced in Sec- where N (t ) is the total normal force, Npre is the pre-compressor
tion 2. Section 3 gives a brief overview of FOPID controller. An force, CPZ is the friction coefficient of the PFD and V (t ) is the ap-
overview of MOCS is presented in Section 4. The FOPID controller plied voltage to the stack actuator. Therefore, the friction force of
is designed for a smart base-isolated building in Section 5. Nu- the semi-active damper is given by
merical studies are given in Section 6. Moreover, the performance u (t ) = μd N (t ) Sgn (ẋ) if ẋ ≠ 0 (6)
of the proposed controller is compared with that of several control
strategies in this section. The concluding remarks are summarized
in Section 7. −μd N (t ) < u (t ) < μd N (t ) if ẋ = 0 (7)

where μd is the friction coefficient of materials between the fric-

tion bar and friction pad and Sgn (x ̇) is the sign of the slip velocity
2. Modeling of a base-isolated structure with PFDs
of the damper.
The control forces obtained from each of the linear feedback
The equation of motion of an n-degree-of-freedom linear base-
control laws are active control forces which cannot be achieved by
isolated structure, subjected to earthquake ground acceleration
semi-active devices. The PFD in PSIS only produces a resistance
x¨ g (t ), is given by force and its direction is always opposite to that of the slip velo-
Mx¨ (t ) + Cẋ (t ) + Kx (t ) = − Mrx¨ g (t ) (1) city. Also, the desired control force can be realized when the
damper is in the slipping phase [7]. In this case, the following
where M, C and K are n × n mass, damping and stiffness matrices, equation is used to adjust the desired control force, N (t ).
respectively. x (t ), x ̇ (t ) and x¨ (t ) denote displacement, velocity and
⎧ uActive
acceleration vectors, respectively. Also, r is the location vector for ⎪ if uActive ẋb < 0
an earthquake. When control force, u (t ) are applied to the struc- N (t ) = ⎨ μ
⎪
ture, the equation of motion can be written as ⎩0 if uActive ẋb ≥ 0 (8)

Mx¨ (t ) + Cẋ (t ) + Kx (t ) = − Mrx¨ g (t ) + Du (t ) (2) where uActive is the control force obtained from the linear feedback
control method and μ is Coulomb friction coefficient. After de-
where D and u (t ) represent an n × nc location matrix of control termining contact force N (t ), the actuator voltage V (t ), which is an
forces and nc × 1 control force vector for nc actuators, respectively. online control law, can be directly determined from Eq. (9).
Considering z (t ) = [ x (t ) x ̇ (t )]T , Eq. (2) in state-space form gives
N (t )−No
z ̇ (t ) = Az (t ) + Bu (t ) + Hx¨ g (t ) (3) 0 ≤ V (t ) = ≤ Vmax
CPZ (9)
where the state matrix, A, and input matrix, B and H are as follows. To evaluate the FOPID performance in reduction of the struc-
⎡ ⎤ tural responses under earthquake loads, a five-story base-isolated
0 I
A=⎢ ⎥, building, studied in [36], is taken into account. Considering one
⎣ − M−1K − M−1C⎦
degree of freedom for the base story and 5 degrees of freedom for
⎡ 0 ⎤
B=⎢ ⎥, the superstructure stories, a lumped-mass structural model with
⎣ M−1D⎦ six degrees of freedom is used for the numerical simulation. The
⎡ ⎤ mass of each story is equal to 5897 kg. The damping coefficients
H=⎢ 0⎥
⎣ − r⎦ (4) for the five floors are 67, 58, 57, 50, and 38 kNs/m, respectively. The
corresponding stiffness coefficients are 33732, 29093, 28621,
In Eq. (4), I and 0 are the identity and zero matrices,
24954, and 19059kN/m, respectively. The fundamental period and
respectively. damping ratio of the structure without isolator are 0.3 s and 0.02,
Fig. 1 shows a schematic diagram of the PSIS components. A respectively. The isolation system, which consists of low-damping
PSIS consists of an isolation system and PFDs. During an earth- rubber bearings, assumed to have a linear force deformation be-
quake excitation, the relative motion between the friction bar and havior with a viscous damping. The total mass, damping coeffi-
friction pad will generate a sliding force. A piezoelectric actuator, cient and stiffness of the rubber bearings are 6800 kg, 7.48 kNs/m
embedded in the PFD, provides a controllable normal force be- and 231.5kN/m, respectively. The base-isolated structure has a
tween the friction pad and friction bar. The normal (contact) force period of 2.5 s and a damping ratio of 0.04 in the first mode. The
provided by the PFD can be regulated by changing the input vol- PFDs with a total force capacity of 16.8kN are installed on the base
tage of the piezoelectric actuator. A pair of wedge blocks is used to of the structure. The PFD parameters are the parameters used in
pre-compress the piezoelectric actuator. Furthermore, a load cell is experimental studies by Lu and Lin [11]. The parameters Npre , CPZ
embedded in the PFD to measure the vertical force [11,12]. The and μd are 1000 N, 1.10 N/V and 0.2, respectively. Also, the max-
normal force, which is applied to the friction interface of the PFD, imum actuator voltage is 1000 V. The modeling validation for the
is given by PFD has been carried out in Lu and Lin [12].

Fig. 1. The PSIS components: (a) side view, (b) top view [11].
 A.-A. Zamani et al. / ISA Transactions 67 (2017) 222–232 225

4. Multi-objective cuckoo search algorithm

Metaheuristic algorithms such as genetic algorithm, particle

swarm optimization, harmony search, and cuckoo search can ef-
ficiently handle complex optimization problems. To generate new
candidate solutions, the CS [41] applies Lévy flights. Having in-
finite mean and variance, Lévy flights can explore the search space
better than standard Gaussian processes [42]. Also, they are more
efficient than simple random walks. Due to the advantages of the
CS in terms of ease of implementation, few parameters to tune and
fast convergence rate, it provides a better performance in com-
Fig. 2. Expansion of point to plane in FOPID controller. passion with a number of existing algorithms, such as PSO and GA
[43,44].
3. Fractional order PID controllers The population size n, switching probability pa , step-size α and
Lévy flights exponent β are configurable parameters of the CS.
In order to control dynamical systems, fractional order con- Apart from the population size, the switching probability is the
trollers were introduced by Oustaloup [37–39]. He showed the key parameter of the CS. Other parameters such as the step-size
superior performance of this controller over the classical PID and Lévy flights exponent can be set as constant numbers. In
controller. Later, Podlubny presented FOPID controllers as a gen- contrast with α and β , which have constant values of 0.1 and
eralized form of standard PID controllers [40]. The transfer func- 1.5 for most problems, pa and n are variable parameters and have
tion of the FOPID controller, also known as PI λD μ , is as follows great effects on the algorithm performance. The switching prob-
ability provides a balance between the local and global optimiza-
KI tion. With increasing pa , the probability of global optimization is
C (s ) = KP + + KD s μ, (λ , μ > 0) reduced and vice versa [41–44].
sλ (10)
For optimization problems with many conflicting objectives,
where KP , KI and KD are proportional, integral and derivative gains, Yang and Deb [45] presented an MOCS algorithm. The perfor-
respectively. Also, λ and μ are, respectively the order of integral mance of MOCS algorithm was compared with several well-known
and derivative operators. The control signal is given by multi-objective optimization algorithms such as NSGA-2 in [45]. It
was shown that the convergence of MOCS was the highest and it
U (s ) = KP E (s ) + KI s −λE (s ) + KD s μE (s ) (11) led to better solutions in a more efficient way. In the MOCS, the
following rules are used for a problem with K objectives.
where E (s ) and U (s ) are the error and control signals, respectively.
As the FOPID controller has two extra parameters in comparison 1. Each cuckoo lays K eggs at a time, and dumps them in a ran-
with conventional PID controller, its tuning can be much more domly chosen nest. Egg k represents the solution for the kth
complex. objective.
Fig. 2 illustrates the relationship between FOPID and standard 2. The best nests having high quality of eggs will be transferred to
PID controllers. In fact, P, PD, PI, and PID controllers are special the next generations.
3. Each nest will be abandoned with a probability pa and a new
cases of FOPID controllers. According to this figure, FOPID con-
nest with K eggs will be built, according to the similarities/dif-
troller maps the integrative and derivative orders of standard PID
ferences of the eggs.
controller from one point to the λ − μ plane. This expansion pro-
vides more flexibility to the control design problem. Moreover, The first rule simulates a randomization process. A new solu-
FOPID controllers can perform better than standard PID controllers tion can be randomly generated either by a random walk or by a
in term of control requirements, namely the setpoint regulation, Lévy flight. The first rule also performs a localized random per-
disturbance rejection and robustness [34,35]. Based on these ad- mutation and can be considered as a form of crossover. For each
vantages, we suggest to employ FOPID controllers for the seismic nest, there can be K solutions, which are generated according to
control of structures. Eq. (12). To ensure the proper convergence of the algorithm, the

Fig. 3. The basic steps of the MOCS.

226 A.-A. Zamani et al. / ISA Transactions 67 (2017) 222–232

best solutions are transferred to the next generation using the min [f1 (x), f2 (x)]
X
second rule. Hence, the second rule corresponds with elitism. In
the third step, the worst solutions are thrown away with a prob- subject to g (x) ≤ b (18)
ability and new solutions are generated. Therefore, it can play the maxt ‖xbase (t )‖ maxt ‖afloor (t )‖
where f1 (x ) = , f2 (x ) = and ‖. ‖ indicate the
role of mutation. Based on these three rules, the basic steps of the max ‖x^
t (t )‖
base max ‖a^
t (t )‖
floor

MOCS can be summarized as the pseudo code shown in Fig. 3. maximum base displacement and peak floor accelerations in the
For cuckoo i, the new solutions, x it + 1, can be generated using controlled structure normalized by its corresponding values in the
Lévy flight as shown in Eq. (12). uncontrolled structure, and the vector magnitude, respectively.
The uncontrolled structure is a structure with no control force and
xit + 1 = xit + α ⊕ L évy (β ) (12) control tools. As f1 (x ) and f2 (x ) are in conflict with each other, the
Pareto front of the multi-objective optimization procedure pro-
The product ⊕ means entry wise multiplications and α is given vides the possibility to choose a good solution from the presented
by Eq. (13). solutions. In Eq. (18), g (x ) ≤ b is a constraint referring to the lower
α = α0 (xtj − xit ) and upper bounds of the FOPID parameters. In this study, these
(13)
bounds are set to 0 < KP ≤ 200,0 < KI ≤ 200 ,0 < KD ≤ 200, 0 ≤ λ ≤ 2
where α0 is a constant and xtj , x it
are two different randomly se- and 0 ≤ μ ≤ 2. Also, the simulation time is 50 seconds.
lected solutions. Using Lévy flights, the step size s is generated as To design an FOPID controller for seismic control of a smart
shown in Eq. (14). base-isolated structure equipped with PFDs system, the integer-
order approximation of FOPID transfer function must be calculated
u
s = α0 (xtj − xit ) ⊕ L évy (β ) = 0.01 1
(xtj − xit ) using Oustaloup method. The parameters used in Oustaloup ap-
νβ (14) proximation are ωl = 0.01, ωh = 100 and N = 6.
To model earthquakes, an artificial ground acceleration is used.
In addition, the parameters u and ν are given by the normal
It is produced by applying a Gaussian white noise to a filter.
distributions as shown in Eq. (15).
Considering several earthquakes, Nagarajaiah and Narasimhan
u = N (0, σu2 ), ν = N (0, σν2 ) (15) [47] introduced a modified form of the Kanai–Tajimi filter as
shown in Eq. (19).
where σu and σν are as follows.
4ξg ωg s
⎧ Γ (1 + β ) sin (πβ/2) ⎫1/ β F (s ) =
s2 + 2ξg ωg s + ω2 g (19)
σu = ⎨ ⎬ , σν = 1
⎩ Γ [(1 + β )/2] β2(β − 1) /2 ⎭ (16)
where ξg and ωg are the ground damping and frequency, respec-
Eq. (17) shows how a new solution, x it + 1, is generated using a tively. In this study, ξg = 0.3 and ωg = 2π rad/s are used in the
random walk numerical simulations. The output of this filter simulates the
earthquake.
xit + 1 = xit + αs ⊕ H (pa − ε) ⊕ (xtj − xkt ) (17) The block diagram of MOCS-FOPID for smart base-isolated
structures equipped with PFDs is illustrated in Fig. 4.
where, H (u) is the Heaviside function, s > 0 is the step size scaling
In this research work, the values of the population size and
factor, and ε is a random number with uniform distribution [46].
switching probability are 20 and 0.25, respectively. The time his-
tory analysis of the base-isolated building equipped with PFDs
subjected to the artificial earthquake is carried out in MATLAB
5. Semi-active control of base-isolated structures equipped [48]. The parameters of the FOPID controller with different feed-
with PFDs back signals are determined using a MOCS optimizer. Fig. 5 shows
the set of Pareto optimal solutions for the isolation displacement,
The effectiveness of an FOPID controller essentially depends on isolation velocity and roof acceleration as different feedback sig-
the values of the proportional, integral and derivative coefficients nals. It is worth mentioning that each member of the Pareto op-
as well as the order of the integral and derivative operators. Due to timal front is a possible solution. Considering both objective
the uncertainties in the external excitation and structural system, functions, Fig. 5 shows that the best solutions are achieved when
however, it is a challenging task to determine the suitable con- the feedback signal is the isolation velocity. Considering this
feedback signal, the selected solution should make a trade-off
troller parameters. In addition, the large displacement of the iso-
between the conflicting design objectives. In other words, an FO-
lator in near-field motions is a major challenge. The PFD, as a VFD
PID controller minimizing the isolation displacement subjected to
device, can be used to overcome this issue. For semi-active control
no significant increase in the acceleration response of the
of a base-isolated building using PFD, however, sudden changes in
the damper friction force of PFDs can increase the floor accelera-
tions of superstructure. To avoid the increase of the floor accel-
erations of superstructure, a suitable control system is required to
reduce the isolation displacement without any increase in the roof
acceleration during various earthquake excitations.
Design of an FOPID controller can be seen as an optimization
problem. In fact, the main issue is to determine
x = ⎡⎣ KP , KI , KD, λ , μ⎤⎦ minimizing the conflicting performance in-
T

dices. To solve an optimization problem with conflicting design

objectives, a prompt solution is to use a multi-objective evolu-
tionary algorithm. To obtain a good trade-off between the isolation
displacement and superstructure acceleration using FOPID control, Fig. 4. The block diagram of MOCS-FOPID for smart base-isolated structures
the design problem is formulated as follows equipped with PFDs.
 A.-A. Zamani et al. / ISA Transactions 67 (2017) 222–232 227

2.6 Using MATLAB, the values of performance indices are given by

MOCS-FOPID with isolation velocity feedback Tables 2 and 3 for the components of FP and FN of real-data
2.4 MOCS-FOPID with isolation displacement feedback
earthquakes. Also, the values of nine performance indices given by
MOCS-FOPID with roof acceleration feedback
2.2 the proposed controllers are compared with those provided by
other techniques, i.e. maximum passive operation of PFDs, mod-
2 ified clipped-optimal controller [14], supervisory fuzzy logic con-
troller [13], GA-based self-organizing fuzzy logic controller [13],
1.8
optimal PID controller [3] and genetic algorithm hierarchic fuzzy
f2(x)

1.6 logic control [15].

In most considered earthquakes, it can be seen that the
1.4 maximum passive operation of PFDs and the proposed controller
significantly reduce J3 and J7, which are performance indices re-
1.2
lated to the maximum base displacement and the value of its
1 RMS, respectively. Considering indices J5 and J8, it can be seen
that the maximum passive operation results in a significant in-
0.8 Selected crease in the maximum floor acceleration and the value of its
solution
RMS in the most cases. For example, in comparison with the
0.4 0.6 0.8 1 1.2 1.4 uncontrolled structure, the maximum passive operation of PFD
f1(x) resulted in 73%, 80% and 38% increase in peak story acceleration
for the FN component of El Centro, Newhall and Kobe earth-
Fig. 5. Set of Pareto optimal solutions for MOCS-FOPID using different feedbacks.
quakes, respectively. Considering these earthquakes, this increase
is as much as 131%, 135%, and 156% for the FP component, re-
spectively. At the cost of deterioration in the peak isolation de-
superstructure is required. For the selected solution, the FOPID is formation, in comparison with the maximum passive operation,
given by Eq. (20). the modified clipped-optimal controller performs better in re-
duction of the maximum floor acceleration. Compared to the
159.5
C (s ) = 185.4 + + 184.8s 0.26 uncontrolled cases, for the FN component of El Centro, Newhall
s 0.13 (20)
and Kobe earthquakes, for example, the modified clipped-optimal
controller results in 71%, 30% and 20% increase in the maximum
6. Results and discussions accelerations of floors, respectively. These increases are 80%, 59%,
and 95% for the FP component of these earthquakes, respectively.
To evaluate the performance of the proposed controller, a set of Unlike the maximum passive operation and the modified clip-
fault-parallel and fault-normal components of El Centro, Newhall, ped-optimal controller, the SVFLC, SOFLC, GHFLC, and PID con-
Sylmar, Rinaldi, Kobe, Chi-Chi, Erzinkan, Considered by Narasimhan trollers successfully limit the increase in the peak floor accel-
et al. [2], are used. The simulation results are evaluated by a set of eration in most earthquake excitation, while reducing the max-
nine performance indices, J1 to J9, as shown in Table 1 [2]. The imum isolator deformations satisfactorily. For some earthquakes,
maximum base shear in the controlled structure normalized by its however, the results from SVFLC, SOFLC, GHFLC indicate that any
corresponding value in the uncontrolled structure is shown by J1. attempt to decrease the maximum base displacement leads to
Similarly, the structural shear, base displacement, story drift and increasing the superstructure accelerations in comparison with
floor acceleration in the controlled structure normalized by their the uncontrolled case. Considering J3 and J7, which are perfor-
corresponding values in the uncontrolled structure are represented mance indices related to the maximum base displacement and
by indices J2 to J5, respectively. J6 represents the maximum force the value of its RMS, it can be seen that the MOCS-FOPID con-
generated by the PFD normalized by the maximum base shear in troller is able to significantly reduce the maximum base dis-
the controlled structure. Indices J7 and J8 determine the root mean placement and the value of its RMS. Furthermore, by comparing
square of the base displacement and floor acceleration of the con- the results provided by the MOCS-FOPID controller with those
trolled structure normalized to their values in the uncontrolled given by other control techniques in reduction of performance
structure, respectively. Finally, J9 represents the energy absorbed by indices J5 and J8, the maximum floor acceleration and the value of
the control tool in the controlled structure normalized by the its RMS, it can be concluded that the proposed controller pro-
earthquake input energy in the controlled structure. vides the best performance.
Compared with maximum passive operation, for example, the
Table 1 MOCS-FOPID controller reduces the maximum floor acceleration for
Performance indices.
the FP component of the El Centro, Newhall and Kobe earthquakes
Peak base sheara Peak inter story drifta RMS base displacement
by 70%, 65% and 69%, respectively. Similarly, this reduction is as
maxt ‖V0 (t )‖ maxt , f ‖df (t )‖ maxi‖σd (t )‖ much as 56%, 56% and 33%, for the FN component of the same
J1 = J4 = J7 =
^
maxt ‖V0 (t )‖ ^ maxi‖σ^d (t )‖ earthquakes, respectively. Considering the FP component of El
maxt , f ‖df (t )‖

Peak structural shear Peak floor acceleration RMS floors accelerationa Centro earthquake, there are reductions of 54%, 48%, 55% and 26% for
J2 =
maxt ‖V1(t )‖
J5 =
maxt , f ‖af (t )‖
J8 =
maxf ‖σa (t )‖ GHFLC, SVFLC, SOFLC and PID controller in the maximum base dis-
^
maxt ‖V1(t )‖ maxt , f ‖a^f (t )‖ maxf ‖σ ^ (t )‖
a placement in comparison with uncontrolled structure. This reduc-
Peak base displacement Peak control force Energy dissipated by PFD tion is 75% for MOCS-FOPID controller. For the FN component of
maxt ‖xb (t )‖ maxt ‖ud (t )‖ T
J3 =
maxt ‖x^b (t )‖
J6 =
maxt ‖V0 (t )‖ J9 =
∫0 ud (t ) xḃ (t ) dt Chichi earthquake, these reductions are 39%, 38%, 36%, 32% and 46%
T
∫0 〈V0 (t ) xġ (t )〉dt for GHFLC, SVFLC, SOFLC, PID controller and MOCS-FOPID controller,
respectively. Compared to the uncontrolled cases for the FP com-
a
The denominator consists of the corresponding response quantity in the ponent of El Centro and Kobe earthquakes, SVFLC leads to an in-
uncontrolled case f ¼ floor number, 1,…, 5; t ¼ time; ‖. ‖ ¼ vector magnitude; 〈〉 ¼
inner product;V0, V1 ¼ base and structural shears; xb ¼ base displacement; df ¼ inter
crease of 33% and 34% in terms of the maximum floor acceleration.
story drift; af ¼ floor acceleration; σd and σa ¼ RMS base displacement and floor Similarly, these increases are 62% and 39% for SOFLC and 25% and
acceleration; ^ ¼ corresponding response quantity in the uncontrolled case. 21% for GHFLC, respectively. However, PID and MOCS-FOPID
228 A.-A. Zamani et al. / ISA Transactions 67 (2017) 222–232

Table 2
Performance indices of the base-isolated building subjected to FP component of earthquakes.

Earthquake Performance Indexes J1 J2 J3 J4 J5 J6 J7 J8 J9

El Centro Maximum Passive operation 0.60 0.78 0.24 1.00 2.31 0.59 0.16 1.60 0.93
Modified clipped optimal controller 0.68 0.76 0.51 0.76 1.80 0.52 0.33 1.28 0.87
Optimal PID 0.78 0.78 0.74 0.79 0.79 0.15 0.60 0.66 0.51
SVFLC 0.69 0.69 0.52 0.69 1.33 0.35 0.33 0.92 0.90
SOFLC 0.63 0.66 0.45 0.66 1.62 0.45 0.26 0.98 0.93
GHFLC 0.64 0.68 0.46 0.71 1.25 0.45 0.25 0.96 0.95
MOCS-FOPID 0.66 0.69 0.25 0.80 0.69 0.47 0.16 0.69 0.95

Newhall Maximum Passive operation 0.53 0.74 0.30 1.00 2.35 0.42 0.17 0.95 0.91
Modified clipped optimal controller 0.55 0.64 0.44 0.71 1.59 0.40 0.37 0.75 0.80
Optimal PID 0.87 0.87 0.59 0.74 0.90 0.12 0.55 0.67 0.47
SVFLC 0.50 0.57 0.38 0.64 1.26 0.36 0.32 0.59 0.85
SOFLC 0.51 0.57 0.36 0.67 1.34 0.35 0.24 0.59 0.90
GHFLC 0.49 0.59 0.37 0.68 1.24 0.36 0.31 0.58 0.87
MOCS-FOPID 0.78 0.77 0.50 0.75 0.83 0.18 0.30 0.62 0.51

Sylmar Maximum Passive operation 0.62 0.65 0.50 0.65 0.77 0.19 0.29 0.50 0.82
Modified clipped optimal controller 0.77 0.78 0.72 0.78 0.90 0.15 0.51 0.60 0.62
Optimal PID 0.73 0.73 0.62 0.67 0.74 0.20 0.44 0.49 0.70
SVFLC 0.71 0.73 0.65 0.73 0.82 0.15 0.44 0.50 0.71
SOFLC 0.68 0.71 0.60 0.71 0.82 0.17 0.38 0.47 0.76
GHFLC 0.69 0.70 0.66 0.70 0.81 0.14 0.30 0.45 0.73
MOCS-FOPID 0.54 0.57 0.51 0.62 0.62 0.19 0.30 0.36 0.74

Rinaldi Maximum Passive operation 0.62 0.63 0.52 0.63 0.72 0.16 0.34 0.50 0.80
Modified clipped optimal controller 0.81 0.81 0.76 0.81 0.97 0.12 0.56 0.61 0.58
Optimal PID 0.62 0.62 0.62 0.66 0.63 0.21 0.44 0.48 0.67
SVFLC 0.74 0.73 0.68 0.73 0.78 0.12 0.47 0.53 0.69
SOFLC 0.71 0.71 0.64 0.71 0.77 0.14 0.44 0.51 0.72
GHFLC 0.69 0.72 0.62 0.70 0.71 0.12 0.42 0.52 0.71
MOCS-FOPID 0.59 0.59 0.51 0.70 0.60 0.16 0.33 0.39 0.86

Kobe Maximum Passive operation 0.75 0.80 0.49 0.94 2.56 0.35 0.43 1.69 0.87
Modified clipped optimal controller 0.90 1.04 0.65 1.12 1.95 0.29 0.62 1.56 0.75
Optimal PID 0.77 0.77 0.65 0.74 0.79 0.19 0.72 0.79 0.50
SVFLC 0.75 0.81 0.59 0.84 1.34 0.32 0.58 1.03 0.80
SOFLC 0.71 0.80 0.59 0.85 1.39 0.36 0.53 1.05 0.83
GHFLC 0.68 0.82 0.55 0.82 1.21 0.35 0.51 1.01 0.81
MOCS-FOPID 0.71 0.74 0.51 0.82 0.79 0.24 0.45 0.70 0.79

Chichi Maximum Passive operation 0.67 0.68 0.55 0.68 0.81 0.18 0.47 0.68 0.56
Modified clipped optimal controller 0.71 0.71 0.65 0.71 0.92 0.17 0.62 0.71 0.43
Optimal PID 0.70 0.70 0.67 0.69 0.69 0.23 0.60 0.63 0.11
SVFLC 0.70 0.70 0.63 0.70 0.73 0.16 0.56 0.64 0.51
SOFLC 0.70 0.70 0.61 0.70 0.75 0.17 0.55 0.67 0.53
GHFLC 0.68 0.67 0.60 0.68 0.73 0.17 0.50 0.64 0.53
MOCS-FOPID 0.68 0.68 0.57 0.66 0.69 0.18 0.48 0.53 0.87

Erzincan Maximum Passive operation 0.76 0.76 0.59 0.79 1.53 0.21 0.34 0.61 0.86
Modified clipped optimal controller 0.86 0.86 0.78 0.86 1.03 0.19 0.51 0.63 0.72
Optimal PID 0.80 0.81 0.75 0.81 0.82 0.18 0.55 0.58 0.50
SVFLC 0.84 0.84 0.73 0.84 0.86 0.15 0.46 0.54 0.78
SOFLC 0.83 0.83 0.71 0.83 0.92 0.18 0.43 0.54 0.80
GHFLC 0.82 0.82 0.69 0.84 0.90 0.18 0.41 0.52 0.77
MOCS-FOPID 0.81 0.81 0.60 0.82 0.83 0.18 0.35 0.43 0.92

controllers provide a reduction of 21% and 31% for the FP component the maximum peak floor acceleration and inter-story drift during
of El Centro earthquake and a reduction of 21% for the FP component earthquake excitations. Hence, the proposed controller can prevent
of Kobe earthquake. These comparison results reveal that the pro- excessive acceleration and inter-story drift being transmitted up to
posed control strategy is able to simultaneously reduce two im- the isolated equipment and can limit the risks of non-structural
portant structural responses, namely the maximum base displace- earthquake damages.
ment and the maximum floor accelerations. Moreover, the values of Considering the values of performance indices J1 and J2, which
J4 show that the proposed controller performs better than other are related to the maximum base shear and structural shear, it can
control strategies in terms of the inter-story drifts, in most earth- be seen that the best results are obtained by the proposed con-
quakes. Non-structural earthquake damages depend on the values of troller in most earthquakes. The values of J6 indicate that the
 A.-A. Zamani et al. / ISA Transactions 67 (2017) 222–232 229

Table 3
Performance indices of the base-isolated building subjected to FN component of earthquakes.

Earthquake Performance Indexes J1 J2 J3 J4 J5 J6 J7 J8 J9

El Centro Maximum Passive operation 0.58 0.59 0.28 0.71 1.73 0.51 0.18 1.59 0.94
Modified clipped optimal controller 0.66 0.73 0.51 0.79 1.71 0.46 0.35 1.25 0.86
Optimal PID 0.89 0.90 0.69 0.77 0.92 0.14 0.57 0.66 0.48
SVFLC 0.62 0.67 0.47 0.68 1.25 0.32 0.33 0.98 0.89
SOFLC 0.60 0.60 0.40 0.66 1.20 0.40 0.26 1.05 0.93
GHFLC 0.64 0.65 0.49 0.69 1.22 0.31 0.31 1.01 0.90
MOCS-FOPID 0.66 0.69 0.29 0.63 0.76 0.40 0.18 0.77 0.85

Newhall Maximum Passive operation 0.95 0.99 0.74 0.99 1.80 0.22 0.62 1.41 0.83
Modified clipped optimal controller 0.97 1.00 0.86 1.00 1.30 0.22 0.77 1.11 0.69
Optimal PID 0.87 0.86 0.86 0.96 0.89 0.14 0.65 0.74 0.45
SVFLC 0.97 0.96 0.83 0.96 1.13 0.20 0.74 1.04 0.76
SOFLC 0.98 1.00 0.81 1.00 1.27 0.22 0.72 1.09 0.78
GHFLC 0.99 0.98 0.82 0.98 1.11 0.21 0.74 1.14 0.77
MOCS-FOPID 0.77 0.73 0.76 0.83 0.80 0.12 0.50 0.71 0.47

Sylmar Maximum Passive operation 0.69 0.71 0.60 0.71 0.77 0.13 0.41 0.51 0.74
Modified clipped optimal controller 0.77 0.78 0.73 0.78 0.89 0.11 0.63 0.66 0.53
Optimal PID 0.95 0.94 0.61 0.65 0.94 0.11 0.44 0.49 0.73
SVFLC 0.72 0.73 0.65 0.73 0.79 0.11 0.52 0.56 0.65
SOFLC 0.70 0.74 0.64 0.74 0.85 0.12 0.47 0.51 0.69
GHFLC 0.71 0.73 0.64 0.75 0.91 0.11 0.50 0.55 0.66
MOCS-FOPID 0.89 0.87 0.60 0.77 0.89 0.07 0.42 0.46 0.51

Rinaldi Maximum Passive operation 0.88 0.91 0.75 0.91 1.07 0.15 0.54 0.86 0.77
Modified clipped optimal controller 0.90 0.92 0.84 0.92 1.05 0.15 0.67 0.89 0.60
Optimal PID 0.95 0.96 0.87 1.01 1.00 0.30 0.67 0.73 0.54
SVFLC 0.93 0.95 0.81 0.95 0.99 0.13 0.63 0.75 0.70
SOFLC 0.91 0.91 0.80 0.91 0.93 0.15 0.60 0.74 0.71
GHFLC 0.90 0.91 0.79 0.92 0.95 0.14 0.60 0.72 0.70
MOCS-FOPID 0.77 0.79 0.74 0.92 0.82 0.15 0.53 0.65 0.69

Kobe Maximum Passive operation 0.77 0.84 0.57 0.86 1.38 0.25 0.59 1.69 0.82
Modified clipped optimal controller 0.77 0.89 0.78 0.89 1.20 0.22 0.73 1.44 0.68
Optimal PID 0.53 0.53 0.50 0.52 0.53 0.19 0.44 0.46 0.21
SVFLC 0.87 0.88 0.74 0.88 0.93 0.21 0.70 1.13 0.75
SOFLC 0.85 0.87 0.72 0.87 1.05 0.23 0.67 1.19 0.77
GHFLC 0.86 0.86 0.75 0.88 1.02 0.23 0.65 1.14 0.76
MOCS-FOPID 0.84 0.83 0.64 0.68 0.92 0.15 0.59 0.81 0.41

Chichi Maximum Passive operation 0.65 0.65 0.59 0.65 0.68 0.09 0.49 0.55 0.63
Modified clipped optimal controller 0.80 0.80 0.77 0.80 0.82 0.07 0.70 0.72 0.42
Optimal PID 0.75 0.75 0.68 0.73 0.75 0.20 0.52 0.55 0.63
SVFLC 0.66 0.66 0.62 0.66 0.70 0.08 0.54 0.57 0.58
SOFLC 0.69 0.69 0.64 0.69 0.78 0.09 0.55 0.58 0.58
GHFLC 0.65 0.63 0.61 0.65 0.71 0.07 0.53 0.59 0.59
MOCS-FOPID 0.60 0.60 0.54 0.60 0.60 0.10 0.50 0.51 0.69

Erzincan Maximum Passive operation 0.80 0.70 0.70 0.80 0.87 0.12 0.51 0.61 0.74
Modified clipped optimal controller 0.85 0.85 0.80 0.85 0.96 0.11 0.68 0.71 0.54
Optimal PID 0.75 0.75 0.70 0.73 0.75 0.18 0.54 0.56 0.61
SVFLC 0.81 0.81 0.74 0.81 0.83 0.11 0.58 0.62 0.66
SOFLC 0.79 0.79 0.72 0.79 0.86 0.12 0.54 0.58 0.71
GHFLC 0.75 0.73 0.70 0.75 0.85 0.12 0.53 0.58 0.69
MOCS-FOPID 0.82 0.82 0.70 0.77 0.86 0.12 0.51 0.54 0.75

proposed controller demands a smaller value of control force in excitation energy. However, as shown in Tables 2 and 3, it usually
most earthquakes while providing a superior performance, in demands the maximum voltage and increases both internal de-
comparison with other control methods. formations and absolute acceleration of the superstructure floors.
The selected control strategy to adjust the contact force of PFDs In other words, its performance in simultaneous reduction of the
plays a key role in the dissipation of input excitation energy and maximum base displacement and maximum acceleration of stor-
seismic performance of base isolated structures. Using J9, the ies is not satisfactory. For most earthquakes, the proposed con-
performance of the proposed controller can be compared with troller is able to dissipate the input excitation energy better than
that of other control approaches in terms of the dissipated energy the modified clipped optimal controller, SVFLC, SOFLC, GHFLC and
by PFDs normalized by the input excitation energy. The maximum optimal PID controller. Also, it performs better than other control
passive operation of piezoelectric friction dampers performs better strategies in terms of simultaneous reduction of the maximum
than other control strategies in terms of dissipation of input base displacement and maximum acceleration of stories.
230 A.-A. Zamani et al. / ISA Transactions 67 (2017) 222–232

Uncontrolled MOCS-FOPID

Isolator disp. (cm)

Isolator disp. (cm)
20 100

0 0

-20 -100
0 5 10 15 20 25 30 0 20 40 60 80
Time (s) Time (s)
Top floor acc. (m/s )

Top floor acc. (m/s )

2

2
2 10

0 0

-2 -10
0 5 10 15 20 25 30 0 20 40 60 80
Time (s) Time (s)

1000 1000
Voltage (V)

Voltage (V)
500 500

0 0
0 5 10 15 20 25 30 0 20 40 60 80
Time (s) Time (s)
(a).El Centro (b).Chichi

Fig. 6. Time responses for uncontrolled and controlled structures using MOCS-FOPID subjected to the (a) FP component of the El Centro and (b) FN component of Chichi
earthquake.

Considering the FP component of the El Centro and FN com- gives a reduction of 40% in comparison with the uncontrolled
ponent of Chichi, as far-field and near-field earthquakes, the time structure. Therefore, the proposed controller is able to decrease
histories of the isolator displacement, top floor acceleration and the base displacement of the isolated building in both far-field and
command voltage are illustrated in Fig. 6. The time histories of near-field earthquakes.
these responses for the uncontrolled structure are also shown in In comparison with other controllers, the proposed controller
this figure. As it shows, in case of a far-field earthquake, such as also shows a better performance in controlling the acceleration
the El Centro excitation, the proposed controllers significantly response. From the maximum voltage point of view, it can be seen
decrease the base displacement of the isolated building. that the proposed controller demands a higher value of voltage in
In case of a far-field earthquake, such as the Chichi earthquake, the near-filed earthquake in comparison with the far-filed one. For
it is observed that the maximum base displacement of the un- the same earthquakes, the hysteresis loops of the isolators for the
controlled structure is about 123 cm while this value is about controlled structure are compared with those of the uncontrolled
74 cm for the controlled structure using MOCS-FOPID. It means building in Fig. 7. The total shear force includes the total friction
that the semi-active control of PFDs using the proposed controller and restoring force in the elastomeric bearings normalized to total

Uncontrolled MOCS-FOPID
0.2 1
Total shear force (%W)

Total shear force (%W)

0.1 0.5

0 0

-0.1 -0.5

-0.2 -1
-20 -10 0 10 20 -200 -100 0 100 200
Isolation displacement (cm) Isolation displacement (cm)
(a).El Centro (b).Chichi
Fig. 7. Hysteresis loops of isolation systems for the uncontrolled and controlled structures using MOCS-FOPID subjected to (a) FP component of the El Centro and (b) FN
component of Chichi earthquake.
 A.-A. Zamani et al. / ISA Transactions 67 (2017) 222–232 231

Nominal stiffness matrix upper bound of stiffness matrix lower bound of stiffness matrix
Elcentro-FN
Elcentro-FP
Rinaldi-FN
Rinaldi-FP
Erzincan-FN
Erzincan-FP
Chich-FN
Chich-FP
Kobe-FN
Kobe-FP
Newhall-FN
Newhall-FP
Sylmar-FN
Sylmar-FP
0 0.5 1 0.6 0.8 1 1.2 1.4 0 0.5 1
J3 J5 J6

Fig. 8. Performance indices of the MOCS-FOPID with uncertainty.

weight of the structure. As shown in this figure, the PFDs increase benchmark building. Part I: problem definition. Struct Control Health Monit
the height and decrease the width of the hysteresis loop. In fact, 2006;13(2–3):573–88.
[3] Etedali S, Sohrabi MR, Tavakoli S. Optimal PD/PID control of smart base-
the proposed controller is able to reduce base displacement of the isolated buildings equipped with piezoelectric friction dampers. Earthq Eng
building by increasing the damping force, while maintaining the Eng Vib 2013;12(1):39–54.
same level of dissipated energy. [4] Chen G, Chen C. Behavior of piezoelectric friction dampers under dynamic
loading. Proceedings of SPIE 2000;vol. 3988:54–63.
To investigate the robustness of the proposed controller to [5] Chen C, Chen G. Shake table tests of a quarter‐scale three‐story building model
model uncertainties, an uncertainty of ±10% in the stiffness is with piezoelectric friction dampers. Struct Control Health Monit 2004;11
considered. For three important performance indices i.e. j3, j5 and j6 , (4):239–57.
[6] Chen G, Chen C. Semiactive control of the 20-story benchmark building with
the performance of the MOCS-FOPID controller is evaluated in Fig. 8.
piezoelectric friction dampers. J Eng Mech 2004;130(4):393–400.
According to this figure, it can be seen that the proposed controller [7] Ng CL, XU YL. Semi-active control of a building complex with variable friction
maintains appropriate performance in dealing with model un- dampers. Eng Struct 2007;29(6):1209–25.
certainties. This means that it is not sensitive to modeling errors and [8] Xu YL, Chen B. Integrated vibration control and health monitoring of building
structures using semi-active friction dampers: part I-methodology. Eng Struct
shows a robust performance against model uncertainties. 2008;30(7):1789–801.
[9] Song G, Sethi V, Li HN. Vibration control of civil structures using piezoceramic
7. Conclusions smart materials: a review. Eng Struct 2006;28(11):1513–24.
[10] Cheng FY, Jiang H, Lou K. Smart structures: innovative systems for seismic
response control. Boca Raton: CRC Press; 2008.
This paper aimed to design an FOPID controller to adjust the [11] Lu LY, Lin GL. A theoretical study on piezoelectric smart isolation system for
contact force of PFDs for semi-active control of base-isolated seismic protection of equipment in near-fault areas. J Intell Mater Syst Struct
structures. To achieve the control objectives, reduction of the 2009;20(2):217–32.
[12] Lu LY, Lin CC, Lin GL, Lin CY. Experiment and analysis of a fuzzy-controlled
isolation system deformation without notable increase of the su- piezoelectric seismic isolation system. J Sound Vib 2010;329(11):1992–2014.
perstructure accelerations, a MOCS algorithm was employed to [13] Ozbulut OE, Hurlebaus S. Fuzzy control of piezoelectric friction dampers for
tune the FOPID parameters. Considering the resulting Pareto op- seismic protection of smart base-isolated buildings. Bull Earthq Eng 2010;8
(6):1435–55.
timal curve, the base velocity of the structure was selected as the [14] Ozbulut OE, Bitaraf M, Hurlebaus S. Adaptive control of base-isolated struc-
best input for the FOPID controller. By comparing the performance tures against near-field earthquakes using variable friction dampers. Eng
of the proposed controller with those provided by several control Struct 2011;33(12):3143–54.
[15] Zhao D, Li Y. Fuzzy control for seismic protection of semiactive base-iso-
techniques, it was revealed that the proposed controller had a lated structures subjected to near-fault earthquakes. Math Probl Eng 2015.
better performance in terms of simultaneous reduction of max- http://dx.doi.org/10.1155/2015/675698.
imum base displacement and story acceleration for various types [16] Aguirre N, Ikhouane F, Rodellar J. PI semi active control using MR dampers.
Proceedings of MOVIC2010, 2010.
of earthquakes. Moreover, it provided acceptable responses in
[17] Subasri R, Natarajan AM, Sundaram S. Neural aided discrete PID active con-
terms of inter-story drifts, RMS of base displacements and peak troller for non-linear hysteretic base-isolation building. In: Proceedings of the
floor acceleration. Finally, the robustness of the proposed con- 9th Asian control conference, 2013. p. 1–8.
[18] Nigdeli SM. Effect of feedback on PID controlled active structures under
trollers was assessed by considering a stiffness uncertainty of
earthquake excitations. Earthq Struct 2014;6(2):217–35.
±10% and it was shown that the FOPID controller could provide [19] Yu W, Thenozhi S, Li X. Stable active vibration control system for building
an acceptable performance in dealing with the given model structures using PD/PID control. IFAC Proceedings 2014;vol. 47:4760–5.
uncertainty. [20] Etedali S, Sohrabi MR, Tavakoli S. An independent robust modal PID control
approach for seismic control of buildings. J Civil Eng Urban 2013;3(5):279–91.
[21] Wang ZH, Zheng YG. The optimal form of the fractional-order difference
feedbacks in enhancing the stability of a sdof vibration system. J Sound Vib
References 2009;326(3):476–88.
[22] Muresan CI, Dulf EH, Prodan O. A fractional order controller for seismic mi-
tigation of structures equipped with viscoelastic mass dampers. J Vib Control
[1] Naeim F, Kelly JM. Design of seismic isolated structures: from theory to 2014;2(8):1980–92.
practice. John Wiley & Sons; 1999. [23] Vu TNL, Lee M. Analytical design of fractional-order proportional-integral
[2] Narasimhan S, Nagarajaiah S, Johnson EA, Gavin HP. Smart base‐isolated controllers for time-delay processes. ISA Trans 2013;52(5):583–91.
232 A.-A. Zamani et al. / ISA Transactions 67 (2017) 222–232

[24] Bouafoura MK, Braiek NB. PIλDμ controller design for integer and fractional [35] Koo B, Kwon W, Won S. Active vibration control of a strip in a continuous
plants using piecewise orthogonal functions. Commun Nonlinear Sci Numer galvanizing line using PIλDμ-type fractional-order PID controller. Proceedings
Simul 2010;15(5):1267–78. of SICE, 2015. p. 740–5.
[25] Padula F, Visioli A. Tuning rules for optimal PID and fractional-order PID [36] Johnson EA, Ramallo JC, Spencer BF, Sain MK. Intelligent base isolation sys-
controllers. J Process Control 2011;21(1):69–81. tems. Proceedings of the second world conference on structural control, vol. 1,
[26] Gao Z, Yan M, Wei J. Robust stabilizing regions of fractional-order PDμ con- 1998. p. 367-76.
trollers of time-delay fractional-order systems. J Process Control 2014;24 [37] Oustaloup A. La commande CRONE. Paris: Hermès; 1991.
(1):37–47. [38] Oustaloup A, Moreau X, Nouillant M. The CRONE suspension. Control Eng
[27] Zamani M, Karimi-Ghartemani M, Sadati N, Parniani M. Design of a fractional Pract 1996;4(8):1101–8.
order PID controller for an AVR using particle swarm optimization. Con- [39] Oustaloup A, Sabatier J, Moreau X. From fractal robustness to the CRONE
trol Eng Pract 2009;17(12):1380–7. approach. Proceedings of ESAIM 1998;vol. 5:177–92.
[28] Lee CH, Chang FK. Fractional-order PID controller optimization via improved [40] Podlubny I. Fractional-order system and PIλDμ controllers. IEEE Trans
electromagnetism-like algorithm. Exp Syst Appl 2010;37(12):8871–8.
Autom Control 1999;44(1):208–14.
[29] Gozde H, Taplamacioglu MC. Comparative performance analysis of artificial
[41] Yang XS, Deb S. Cuckoo search via Lévy flights. In: Proceedings of the world
bee colony algorithm for automatic voltage regulator (AVR) system. J Frankl
congress on nature & biologically inspired computing, 2009. p. 210–14.
Inst 2011;348(8):1927–46.
[42] Rajabioun R. Cuckoo Optimization algorithm. Appl Soft Comput 2011;11
[30] Tang Y, Cui M, Hua C, Li L, Yang Y. Optimum design of fractional order PIλDμ
(8):5508–18.
controller for AVR system using chaotic ant swarm. Expert Syst Appl 2012;39
[43] Gandomi AH, Talatahari S, Yang XS, Deb S. Design optimization of truss
(8):6887–96.
structures using cuckoo search algorithm. Struct Des Tall Spec Build 2013;22
[31] Remezanian H, Balochian S, Zare A. Design of optimal fractional-order PID
controllers using particle swarm optimization algorithm for automatic voltage (17):1330–49.
regulator (AVR) system. J Control Autom Electr Syst 2013;24(5):601–11. [44] Valian E, Tavakoli S, Mohanna S, Haghi A. Improved cuckoo search for relia-
[32] Chen Z, Yuan X, Ji B, Wang P, Tian H. Design of a fractional order PID controller bility optimization problems. Comput Ind Eng 2013;64(1):459–68.
for hydraulic turbine regulating system using chaotic non-dominated sorting [45] Yang XS, Deb S. Multi objective cuckoo search for design optimization. Comput
genetic algorithm II. Energy Convers Manag 2014;84:390–404. Oper Res 2013;40(6):1616–24.
[33] Sangpet T, Kuntanapreeda S, Schmidt R. Improving delay-margin of non- [46] Yang XS, Deb S. Cuckoo search: recent advances and applications. Neural
collocated vibration control of piezo-actuated flexible beams via a fractional- Comput Appl 2014;24(1):169–74.
order controller. Shock Vib 2014. http://dx.doi.org/10.1155/2014/809173. [47] Nagarajaiah S, Narasimhan S. Smart base‐isolated benchmark building. part II:
[34] Gad S, Metered H, Bassuiny A, Ghany AMA. Multi-objective genetic algorithm phase I sample controllers for linear isolation systems. Struct Control
fractional-order PID controller for semi-active magnetorheologically damped Health Monit 2006;13(2–3):589–604.
seat suspension. J Vib Control 2015. http://dx.doi.org/10.1177/ [48] MATLAB. The software for numerical computing, version 7.3. The Math Works
1077546315591620. 2006.