Annual Reviews in Control 37 (2013) 346–364

Contents lists available at ScienceDirect

Annual Reviews in Control

journal homepage: www.elsevier.com/locate/arcontrol

Advances in modeling and vibration control of building structures

Suresh Thenozhi, Wen Yu ⇑
Departamento de Control Automatico, CINVESTAV-IPN (National Polytechnic Institute), Mexico City 07360, Mexico

a r t i c l e i n f o a b s t r a c t

Article history: This paper provides an overview of building structure modeling and control. It focuses on different types
Received 18 February 2013 of control devices, control strategies, and sensors used in structural control systems. This paper also dis-
Accepted 14 September 2013 cusses system identification techniques and some important implementation issues, like the time-delay
Available online 11 October 2013
in the system, estimation of velocity and position from acceleration signals, and optimal placement of the
sensors and control devices. Finally, the applications of structural control systems in real buildings and
their performance have been reviewed.
Ó 2013 Elsevier Ltd. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 347
2. Modeling of building structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 347
2.1. Excitation models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 347
2.2. Building structure models. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 348
2.3. Control devices and their models. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 349
2.3.1. Passive devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 349
2.3.2. Active devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 350
2.3.3. Semi-active devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 351
2.3.4. Base isolators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 352
2.3.5. Hybrid devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 352
2.4. Structure-control device models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353
3. Estimation and sensing of structure parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353
3.1. System identification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353
3.2. Sensing and estimation of system states . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 354
4. Control of building structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 355
4.1. Time-delay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 355
4.2. Sensor and actuator placement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 356
4.3. Linear control of building structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 356
4.3.1. PID control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 357
4.3.2. H1 control. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 357
4.3.3. Optimal control. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 357
4.4. Nonlinear control of building structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 357
4.4.1. Sliding mode control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 357
4.5. Intelligent control of building structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 358
4.5.1. Neural network control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 358
4.5.2. Fuzzy logic control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 359
4.5.3. Genetic algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 359
5. Applications of structural vibration control systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 360
6. Conclusions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 361
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 361
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 361

⇑ Corresponding author.
E-mail address: yuw@ctrl.cinvestav.mx (W. Yu).

1367-5788/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.arcontrol.2013.09.012
 S. Thenozhi, W. Yu / Annual Reviews in Control 37 (2013) 346–364 347

1. Introduction time-delay in active control systems and its effects on system sta-
bility is also considered. The paper deals with different control
Protection of large civil structures and human occupants from strategies including the recent Intelligent control techniques like
natural hazards like an earthquake and wind is very important Genetic Algorithm (GA), Neural Network (NN), and Fuzzy logic. Fi-
and challenging. In order to protect buildings, a passive or active nally, the applications of structural control systems in real build-
control is added to the system. Proper formulation and selection ings and their performance have been reviewed. The paper
of the control strategy, mathematical modeling methods, and con- concludes with some of the observations noticed throughout the
trol devices will maximize the effectiveness of the building protec- review.
tion. Due to the importance of the system model in the control
design, a comprehensive review about the building structure and 2. Modeling of building structures
control device modeling is needed.
Recent advances in civil engineering technologies resulted in Structural control concerns mainly with the protection of build-
high-rise buildings. These buildings are sometimes vulnerable to ings from strong winds and seismic loads. In order to control a
natural hazards, which may result in financial, environmental, structure effectively, it is important to have the knowledge about
and human losses. This fact influenced the demand for the protec- its dynamics. A mathematical model of the structure determines
tion of these structures including the human occupants and non- whether a controller is able to produce the desired dynamics in
structural components and systems from natural and man-made the building structure within a stable region (Forrai, Hashimoto,
hazards. One approach to mitigate this undesirable behavior is to Funato, & Kamiyama, 2001; Zhang & Roschke, 1999). The close
alter the dynamic characteristics of the building with respect to a relationship between the control algorithm design and the mathe-
given load. This idea further developed into a new field called matical model is discussed in Housner et al. (1997).
Structural Control, was first presented by Yao in 1972 with a prac-
tical illustration (Yao, 1972). For the past few decades, structural 2.1. Excitation models
control is an active, vast, and growing research area among civil,
mechanical, and control engineers. In order to derive a dynamic model of a building structure, it is
Structural vibration can be generally controlled in two ways: (1) important to know the behavior and impact of the excitations on
by constructing the buildings using smart materials (Housner et al., the buildings, such as strong wind and seismic forces. The force ex-
1997); (2) by adding controlling devices like dampers, isolators, and erted by the earthquake and wind excitation on the structure is
actuators to the building (Balendra, Wang, & Yan, 2001; Chang, shown in Fig. 1. An earthquake is the result of a sudden release
1999; Djajakesukma, Samali, & Nguyen, 2002; McNamara, 1977; of energy in the Earth crust that creates seismic waves. The build-
Yan & Zhou, 2006). In this literature review, we only discuss the lat- ing structure oscillates with the ground motion caused by these
ter case, where the structural dynamics are modified favorably by seismic waves and as a result the structure floor masses experience
adding passive or active devices. The performance of a structural the inertia force. This force can be represented as
control system depends on various factors including excitation type
(e.g., earthquakes and winds), structural characteristics (e.g., degree f ¼ m€xg ð1Þ
of freedom, natural frequency, and structure nonlinearity), control
where m is the mass and €xg is the ground acceleration caused by the
system design (e.g., type and number of devices, placement of de-
earthquake.
vices, system model, and the control algorithm), etc. (Yi & Dyke,
The movement of the structure depends on several factors like
2000). In active control, the structural response under the input
the amplitude and other features of the ground motion, the dy-
excitations are measured using sensors and an appropriate control
namic properties of the structure, the characteristics of the materi-
force, calculated by a pre-assigned controller is used to drive the
als of the structure and its foundation (soil–structure interaction).
actuators for suppressing the unwanted structure vibrations. We
A civil structure will have multiple natural frequencies, which are
will discuss the above mentioned factors in detail in this review.
equal to its number of Degree-of-Freedom (DOF). If the frequency
Due to the popularity and importance of structural control, a
of the motion of the ground is close to the natural frequency of
number of textbooks (Cheng, Jiang, & Lou, 2008; Liang, Lee, Dar-
the building, resonance occurs. As a result, the floors may move
gush, & Song, 2011) and review papers were presented. A brief re-
rigorously in different directions causing inter-story drift, the rela-
view was presented by Housner et al. (1997) in 1997, which
tive translational displacement between two consecutive floors. If
discusses the passive, active, semi-active, and hybrid control sys-
tems and explores the potential of control theory in structural
vibration control. It explains different types of control devices
and sensors used in structural control. The paper concludes with
some recommendations for future research. A recent survey on ac-
tive, semi-active, and hybrid control devices and some control
strategies for smart structures was presented in Fisco and Adeli
(2011, 2011). Some reviews were carried out with particular
emphasis on active control (Datta, 2003; Korkmaz, 2011; Nerves
& Krishnan, 1995; Soong, Masri, & Housner, 1991; Yang & Soong,
1988), on semi-active control (Spencer & Nagarajaiah, 2003), and
on control devices (Soong & Spencer, 2002; Spencer & Sain, 1997;
Symans & Constantinou, 1999).
The aim of this review is to address all aspects involved in struc-
tural control. Compared to the previous reviews, this literature
discusses the mathematical modeling of actuators and structure–
actuator combined system in detail using both linear and nonlinear
methods. This paper also explains the state estimation, system
identification, and optimal device placement techniques. The Fig. 1. (a) Wind excitation. (b) Earthquake excitation.
348 S. Thenozhi, W. Yu / Annual Reviews in Control 37 (2013) 346–364

the building drift value or deformation exceeds its critical point, where M, C, and K 2 Rnn are the mass, damping, and stiffness
the building damages severely. Small buildings are more affected matrices respectively, €xðtÞ; xðtÞ,
_ and xðtÞ 2 Rn1 are the relative
by high-frequency waves, whereas the large structures or high-rise acceleration, velocity, and displacement vectors respectively, and
buildings are more affected by low-frequency waves. The major F 2 Rn1 is the external force vector. The mechanical model of a lin-
part of the structure elastic energy is stored in its low order natural ear n-DOF structure is shown in Fig. 3. The terms in (5) have the fol-
frequencies, so it is important to control the structure from vibrat- lowing definitions,
ing at those frequencies (Choi, Cho, Kim, & Lee, 2005; Jiang, Wei, &
Guo, 2010; Park, Koh, & Seo, 2004). M ¼ diag½ m1 m2    mn 
In the case of high-rise flexible buildings, strong winds cause 2 3
sickness or psychological responses like anxiety to the occupants c1 þ c2 c2  0 0
and also may damage the fragile items. When the vibrations of tal- 6 .. ..7
6 c2 c2 þ c3  . .7
ler buildings due to the high wind exceed a limit of 0.15 m/s2, hu- 6 7
6 .. .. .. .. .. 7
mans may feel uncomfortable (Spencer & Sain, 1997). As a result, C¼6
6 . . . . .
7
7
6 7
the main objective of structural control is to reduce the accelera- 6 .. .. 7
4 . .    cn1 þ cn cn 5
tion response of buildings to a comfortable level. The force exerted
by the wind on a building structure can be represented as Cao 0 0  cn cn
(1997), Li, Liu, Fang, and Tam (2000); 2 3
k1 þ k2 k2  0 0
F w ðhi ; tÞ ¼ !ðhi Þv ðtÞ ð2Þ 6 .. 7
..
6 k2 k2 þ k3  . 7
.
where v(t) is the dynamic wind speed and !(hi) has the following 6 7
6 .. .. .. .. 7
..
expression. K¼6
6 . . . .
7
7.
6 7
!ðhi Þ ¼ qa lp lh Dw ðhi Þv m ð3Þ 6 .. .. 7
4 . .    kn1 þ kn kn 5
where qa is the air density, lp is the wind pressure coefficient, Dw(- 0 0  kn kn
hi) is the windward area of the structure at elevation hi, and vm is The structure displacement under seismic excitation can be referred
the mean wind speed. The wind profile coefficient lh can be ex- in three ways: (a) absolute or total displacement xa(t), (b) ground
pressed as displacement xg(t), and (c) relative displacement x(t) between the
lh ¼ ð0:1hi Þ2aa ð4Þ mass and the ground. The relationship between these three dis-
placements is
where aa is a positive constant.
xa ðtÞ ¼ xg ðtÞ þ xðtÞ ð6Þ
It is worth to note that the main difference between the effects
of earthquake and wind forces on a structure is that, the earth- Thus, the equation of motion governing the relative displacement
quake causes internally generated inertial force due to the building x(t) of the linear structure subjected to ground acceleration €xg ðtÞ is
mass vibration, whereas wind acts in the form of externally applied
m€x þ cx_ þ kx ¼ m€xg ð7Þ
pressure.
Apart from the normal transverse displacement, some models
2.2. Building structure models include the angle of torsion around the centroid of the floor mass.
In this case, the model will contain the angle and inertia parame-
A Single-Degree-of-Freedom (SDOF) structure can be modeled ters and has a transverse-torsional coupled vibration (Saragih,
using three components: the mass component m, the damping 2010). For a structure with n-DOF, the x and M can be modified as
component c, and the stiffness component k (Chopra, 2001), which
is shown in Fig. 2. The stiffness component k can be modeled as x ¼ ½x1 ; . . . ; xn ; h1 ; . . . ; hn T ð8Þ
either a linear or a nonlinear component, in other words elastic M ¼ diag½m1 ; . . . ; mn ; I1 ; . . . ; In  ð9Þ
or inelastic, respectively (Nerves & Krishnan, 1995). Usually the
mass is considered as a constant. When an external force f is ap-
plied to a structure, it produces changes in its displacement x(t),
_
velocity xðtÞ, and acceleration €
xðtÞ.
The stiffness component is said to be linear or elastic, if the rela-
tionship between the lateral force fs and the resulting deformation
is linear (Chopra, 2001). Using Newton’s Second law, the equation
of motion of a linear structure with n-Degree-of-Freedom (n-DOF)
can be expressed as
M€xðtÞ þ C xðtÞ
_ þ KxðtÞ ¼ F ð5Þ

Fig. 2. (a) Structure; (b) stiffness component; (c) damping component; (d) mass
component. Fig. 3. Mechanical model of a n-DOF structure.
 S. Thenozhi, W. Yu / Annual Reviews in Control 37 (2013) 346–364 349

Table 1
Structural control devices (Cheng et al., 2008; Soong & Spencer, 2002; Symans & Constantinou, 1999).

Passive Active Semi-active Isolator Hybrid

TMD, TLCD, Metallic dampers, Friction AMD, MR/ER Dampers, Semi-active TMD, Semi-active TLCD, Elastomeric Bearings, Lead-Plug Bearings, HMD,
dampers, Viscoelastic dampers, Active Friction control devices, Stiffness Control devices, High-Damping Rubber Bearings, Friction HBI
Viscous fluid dampers Tendons Viscous fluid dampers Pendulum Bearings

where hi and Ii are the torsional displacement and inertia of
the ith floor, respectively and T represents the transpose
operation.
Under strong loading demands, the structural behavior is
mostly nonlinear, which can be grouped into geometric nonlinear-
ity and material nonlinearity. In the first case, the geometry of the
structure changes significantly causing the structure to respond
nonlinearly, whereas the stress–strain relationship remains linear.
Material nonlinearities cause a nonlinear stress–strain relation-
ship, which is influenced by the material properties, input load,
and environmental conditions (Kirsch, 2008).
The material nonlinearity can be expressed using the stiffness
matrix. Thus, the equation of motion of a nonlinear structure sub-
jected to ground acceleration €xg ðtÞ is
m€x þ cx_ þ fs ðx; xÞ
_ ¼ m€xg
If the structural elements have plastic or multilinear elastic or
hyper-elastic behavior, then the structural stiffness will change
at different load levels. This time varying behavior of the stiffness
is termed as hysteresis phenomenon, which is amplified under
large deformations (Sain, Sain, & Spencer, 1997). The hysteresis
can be described using different models like the Bouc–Wen model
(Ikhouane & Rodellar, 2007; Spencer, 1986; Wen, 1976), the Hys-
teron (Krasnoselskii & Pokrovskii, 1989), the Chua-Stromsmoe
(Chua & Bass, 1972), and the Preisach models (Brokate & Visintin,
1989; Mayergoyz, 1991). The nonlinear force fs ðx; xÞ _ in (10) can
be modeled using Bouc–Wen model as
_ ¼a
fs ðx; xÞ ~ kx þ ð1  a
~ Þkg
~ fr
In the above expression, fr introduces the nonlinearity, which
satisfies the following condition.
~dx_  m _ r jn~1 fr þ c
~ xjjf
~ðbj _ r jn~ Þ
~xjf
f_ r ¼
g~
where fr is the nonlinear time dependent restoring force,
~ ~ c
d; b; ~; m~; g ~ are the parameters, which controls the shape
~ and n
of the hysteresis loops and system degradation. The variables
~
d; a
~; g~ and k control the initial tangent stiffness (Garrido & Fran-
cisco, 2006).
In the case of n-DOF structures, the nonlinear model can be
modified as
M€xðtÞ þ C xðtÞ
_ þ F s ðxðtÞ; xðtÞÞ
_ ¼ M K€xg ðtÞ
where K 2 R n1
denotes the influence of the excitation force.
2.3. Control devices and their models
The structural vibration control is aimed to prevent struc-
tural damages using vibration control devices. Various control
devices have been developed to ensure the safety of the build-
ing structure even when excessive vibration amplitudes occur
due to earthquake or wind excitations. The control devices are
actuators, isolators, and dampers, which are used to attenuate
the unwanted vibrations in a structure. Many active and passive
devices have been used as vibration control devices. The passive
damper modifies the structure response without using an exter-
nal power supply. Active actuators can generate required forces
for controlling the structure dynamics. Using an external power
supply, these devices will modify the structure stiffness or
damping, which results in a structural dynamics change. The
semi-active device combines the properties of both passive
and active devices. Hybrid devices are formed either by using
both passive and active devices or by using both passive and
semi-active devices. Other well-known vibration control devices
are the base isolators. The list of the commonly used structural
control devices is summarized in Table 1. Basic concepts of
some popular devices are discussed below.
2.3.1. Passive devices
Structural control using passive devices is called passive con-
ð10Þ trol. A passive control device does not require an external power
source for its operation and utilizes the motion of the structure
to develop the control forces. These devices are normally termed
as energy dissipation devices, which are installed on structures
to absorb a significant amount of the seismic or wind induced en-
ergy. The energy is dissipated by producing a relative motion with-
in the control device with respect to the structure motion (Symans
& Constantinou, 1999). For the ideal passive devices, the control
forces applied to the structure are only dependent to the structural
motion, which can be mathematically represented as Yi and Dyke
(2000)
fi ðtÞ ¼ ci x_ di ðtÞ ð14Þ
ð11Þ where x_ di is the relative velocity across the ith device and ci is the
damping coefficient associated with the ith device.
Vibration absorber systems such as Tuned Mass Damper (TMD)
has been widely used for vibration control in mechanical systems.
Basically, a TMD is a device consisting of a mass attached to a
ð12Þ building structure such that it oscillates at the same frequency of
the structure, but with a phase-shift. The mass is usually attached
to the building through a spring-dashpot system and energy is dis-
sipated by the dashpot as relative motion develops between the
mass and structure (Kwok & Samali, 1995). A simple mechanical
model for TMD is depicted in Fig. 4. An early study about the
TMD with a practical application is illustrated in McNamara
(1977).
Tuned Liquid Column Damper (TLCD) dissipates energy similar
to that of TMD, where the secondary mass is replaced with a liquid
ð13Þ column, which results in a highly nonlinear response. They dissi-
pate energy by passing the liquid through the orifices. A simple
mechanical model of TLCD is depicted in Fig. 5. The natural fre-
quency of the TLCD can be obtained as Kim and Adeli (2005)
sffiffiffiffiffiffi
2g
xn ¼ ð15Þ
Lt
where Lt is the length of the liquid tube and g is the acceleration due
to gravity.
The equation of motion of a TLCD satisfies the following expres-
sion (Min, Kim, Lee, Kim, & Ahn, 2005)
1
ql DLt €xv ðtÞ þ ql Dnjx_ v ðtÞjx_ v ðtÞ þ 2ql Dgxv ðtÞ ¼ ql DLh €xðtÞ ð16Þ
2
350 S. Thenozhi, W. Yu / Annual Reviews in Control 37 (2013) 346–364
Fig. 4. Mechanical model of Building-TMD.
Fig. 5. Mechanical model of TLCD.
where xv(t) is the vertical displacement, ql is the liquid density, Lh
and Lt, respectively are the horizontal and total length of the liquid
column, D is the area of cross section, and n is the headless coeffi-
cient. A comparison study of the performance of three types of mass
dampers; TMD, TLCD, and Liquid Column Vibration Absorber (LCVA)
were discussed in Chang (1999) and it is concluded that the TMD
performs better than the other two dampers.
Other passive dampers are (Housner et al., 1997; Soong & Spen-
cer, 2002): metallic yield dampers which dissipate the energy
through the inelastic deformation of metals, friction dampers
which utilize the mechanism of solid friction, develops between
two solid bodies sliding relative to one another, to provide the de-
sired energy dissipation, and viscoelastic dampers that dissipates
the energy through the shear deformation.
Viscous fluid damper works based on the concept of sticky con-
sistency between the solid and liquid. It has a movable piston with-
in a housing filled with highly viscous fluid. The piston contains a
number of orifices, through which the fluid passes from one side to
another that will result in energy dissipation. The output force of
the orifice controlled viscous fluid devices can be expressed as
Kim and Adeli (2005)
f ðtÞ ¼ cjx_ d ðtÞjac sgnðx_ d ðtÞÞ
where x_ d is the relative velocity of the viscous fluid device and ac is
a coefficient in the range of 0.3–2.0.
Passive dampers are very simple and due to the fact that it will
not add energy to the structure, hence it cannot make the structure
unstable. Most of the passive dampers can be tuned only to a par-
ticular structural frequency and damping characteristics. Some-
times, these tuned values will not match with the input
excitation and the corresponding structure response. For example;
(1) nonlinearities. in the structure cause variations in its natural
frequencies and mode shapes during large excitation, (2) a struc-
ture with a Multiple-Degree-of-Freedom (MDOF) moves in many
frequencies during the seismic events. As the passive dampers can-
not adapt to these structure response changes, it cannot assure a
successful vibration suppression (Fisco & Adeli, 2011). This is the
major disadvantage of the passive dampers, which can be over-
come by using multiple passive dampers, each tuned to different
frequencies (e.g., doubly-TMD, Multiple-TMD) or by adding an ac-
tive control to it.
2.3.2. Active devices
The concept of active control has started in early 1970s and the
full-scale application was performed in 1989 (Spencer & Sain,
1997). An active control system can be defined as a system that
typically requires a large power source for the operation of electro-
hydraulic or electromechanical (servo motor) actuator, which in-
creases the structural damping or stiffness. The active control
system uses sensors for measuring both the excitation and struc-
tural responses, and actuators for controlling the unwanted vibra-
tions (Symans & Constantinou, 1999). The working principle of the
active control system is that, based on the measured structural re-
sponse the control algorithm will generate control signal required
to attenuate the vibration. Based on this control signal, the actua-
tors placed in desired locations of the structure generate a second-
ary vibrational response, which reduces the overall structure
response (Soong, 1990). Depending on the size of the building
structure, the power requirements of these actuators vary from
kilowatts to several megawatts (Soong, Reinhorn, Wang, & Lin,
1991). Hence, an actuator capable of generating a required control
force should be used. As the active devices can work with a number
of vibration modes, it is a perfect choice for the MDOF structures. A
number of reviews on active structural control were presented
(Datta, 2003; Housner, Soong, & Masri, 1996; Korkmaz, 2011;
Soong et al., 1991; Yang & Soong, 1988).
The ideal actuators are assumed to have the ability to instanta-
neously and precisely supply force commanded by the control
algorithm (Yi & Dyke, 2000). There are many active control devices
designed for structural control applications. A recent survey on
active control devices is presented in Fisco and Adeli (2011). An Ac-
tive Mass Damper (AMD) or Active Tuned Mass Damper (ATMD) is
created by adding an active control mechanism into the classic
TMD. In this system, 1% of the total building mass is directly ex-
cited by an actuator with no spring and dashpot attached. ATMD
control devices were first introduced in Chang and Soong (1980).
These devices are initially used to reduce structural vibrations
under strong winds and moderate earthquake.
Active tendons are pre-stressed cables, where its stress is con-
trolled using actuators for suppressing the vibration (Fisco & Adeli,
2011). The structural vibration control using active cables and
tendons is presented in Korkmaz (2011). Various numerical
analytical studies have been carried out using tendons for active
control (Alavinasab & Moharrami, 2006; Alli & Yakut, 2005; Kim
& Yun, 2000; Park, Koh, & Ok, 2002; Park et al., 2004; Park, Park, &
Koh, 2008; Singh & Matheu, 1997). At low excitations, the active
ð17Þ control system can be switched-off, then the tendons will resist
the structural deformation in passive mode. At higher excitations,
active mode is switched-on to reach the required tension in
tendons.
A comparison study between active and passive control systems
was carried out in Yi and Dyke (2000) using H2/LQG control algo-
rithm. In simulation, it is found that for SDOF structure both the ac-
tive and passive control systems performed similarly, whereas in
the case of structure with MDOF the active control system showed
high performance.
The active control devices found to be very effective in reducing
the structural response due to high magnitude earthquakes. How-
ever, there are some challenges left to the engineers, such as how
to eliminate the high power requirements, how to reduce the cost
 S. Thenozhi, W. Yu / Annual Reviews in Control 37 (2013) 346–364 351

and maintenance etc. These challenges resulted in the develop- Table 2

ment of semi-active and hybrid control devices (Forrai et al., 2001). MR and ER damper properties (Lord Corporation, 1995; Weiss et al., 1993).

Property MR Damper ER damper

2.3.3. Semi-active devices Max. yield stress 50–100 kPa 2–5 kPa
A semi-active control system typically requires a small external Maximum field 250 kA/m 4 kV/mm
power source for its operation and utilizes the motion of the struc- Plastic viscosity 0.1–1.0 Pa s 0.1–1.0 Pa s
ture to develop control force, where the magnitude of the force can Operable temperature range 40 to 150 °C +10 to 90 °C
Stability Unaffected by Cannot tolerate
be adjusted by an external power source (Symans & Constantinou, most impurities impurities
1999). It uses the advantages of both active and passive devices. Response time Milliseconds Milliseconds
The semi-active devices for structural control application were first Density 3–4 g/cm3 1–2 g/cm3
proposed by Hrovat et al. Hrovat, Barak, and Rabins, 1983. Maximum energy density 0.1 Joules/cm3 0.001 **Joules/cm3
Power supply (typical) 2–25 V; 1–2 A 2000–5000 V; 1–10 mA
The benefits of the semi-active devices over active devices are
their less power requirements, which can even be powered using
a battery that is more important during the seismic events, when
the main power source to the building may fail. Semi-active
devices cannot inject mechanical energy into the controlled struc- modeling tool for both ER and MR dampers. When any field is ap-
tural system, but has properties that can be controlled to optimally plied to these devices, the change in the fluid property can be mod-
reduce the response of the system. Therefore, in contrast to active eled using a Bingham viscoplastic model (Shames & Cozzarelli,
control devices, semi-active control devices do not have the poten- 1992). The plastic viscosity of this model is given in terms of the
tial to destabilize (in bounded-input bounded-output sense) the shear stress and shear strain, which is mathematically represented
structural system (Forrai et al., 2001). A detailed review of semi-ac- as
tive control systems is provided in Spencer and Sain (1997, 2003), s ¼ sy sgnðc_ Þ þ gc_ ð19Þ
Symans and Constantinou (1999), Xu, Agrawal, and Yang (2006).
Like passive friction dampers, these semi-active frictional con- where s is the total shear stress, sy is the yield stress due to the ap-
trol devices dissipate energy through friction caused by the sliding plied field, c_ is the rate of the shear strain, and g is the plastic vis-
between two surfaces. For this damper, a pneumatic actuator is cosity. The relationship between the force and displacement of a MR
provided in order to adjust the clamping force (Pandya, Akbay, damper using this model is given by Xu and Guo (2008)
Uras, & Aktan, 1996). An ideal friction damper can be modeled as 12gN Lp D2p 3Lp sy
a Coulomb element, where the output force is termed as f ¼ _ þ
xðtÞ _
Dp sgn ½xðtÞ ð20Þ
pDi D3p Dp
f ¼ lfn sgnðxÞ
_ ð18Þ
where Lp is the piston length, Dp is the piston cross-sectional area,
where l is the friction coefficient and fn is the normal force (Symans
Di is the inner diameter, Dp is the diameter of the small gap in the
& Constantinou, 1999). In the case of friction dampers, the friction
piston, and gN is the Newtonian viscosity independent of the ap-
coefficient needs to be tuned to have a good energy dissipation. In
plied magnetic field. The yield stress can be represented as a func-
contrast with the passive friction dampers, the semi-active friction
tion of the control current I as follows.
dampers can easily adapt the friction coefficient to varying excita-
tions from weak to strong earthquakes. sy ¼ A1 eI þ A2 lnðI þ eÞ þ A3 I ð21Þ
Semi-active controllable fluid dampers are one of the most
commonly used semi-active control device. For these devices, the where A1, A2 and A3 are the coefficients relative to the MR fluid
piston is the only moving part, which makes them more reliable. property and e is the Euler’s number.
These devices have some special fluid, where its property is mod- Bingham model is a mechanical version of the Bingham
ified by applying external energy field. The electric and magnetic viscoplastic model, which uses damping and Coulomb friction
fields are mainly used to control these devices, which is so called components in the model. This model is further extended, known
as Electro Rheological (ER) and Magneto Rheological (MR) damp- as Gamota and Filisko model, which is a parametric viscoelastic-
ers, respectively (Soong & Spencer, 2002). plastic model. But all of these methods have some shortcomings
ER damper (Symans & Constantinou, 1999): ER dampers consist especially at low velocities. The classic Bouc–Wen can model the
of liquid with micron sized dielectric particles within a hydraulic hysteresis loop pretty well, but fails to predict the roll-off problem
cylinder. When an electric field is applied, these particles will seen at low velocities. A modified Bouc–Wen model was proposed
polarize due to the aligning, thus offers more resistance to flow by Spencer et al. (1997), where an additional damping (c1) and
resulting a solid behavior. This property is used to modify the stiffness (k1) is added to compensate the roll-off and accumulator
dynamics of the structure to which it is attached. stiffness, respectively. The total force of the MR damper is obtained
MR damper (Symans & Constantinou, 1999): The construction as
and functioning of MR dampers are analogous to that of ER damp- ~_ Þ þ k0 ðx  y
f ¼ ab~z þ c0 ðx_  y ~Þ þ k1 ðx  x0 Þ
ers, except the fact that instead of the electric field, magnetic field
is used for controlling the magnetically polarizable fluid. MR ~_ þ k1 ðx  x0 Þ
¼ c1 y ð22Þ
dampers have many advantages over ER dampers, which made
~ and ~z can be found as
where y
them more popular in structural control applications. These de-
vices are able to have a much more yield stress than ER with less ~z_ ¼ cm jx_  y
~_ j~zj~zjn~1  bm ðx_  y
~_ Þj~zjn~ þ dm ðx_  y
~_ Þ ð23Þ
input power. Moreover, these devices are less sensitive to impuri-
1
ties. A comparison between MR and ER fluid dampers is summa- ~_ ¼
y fab ~z þ c0 x_ þ k0 ðx  y ~Þg ð24Þ
c0 þ c1
rized in Table 2.
Different modeling techniques are available to express the where c0 is the viscous damping at large velocities, c1 is the viscous
behavior of these devices, such as; Bingham model, Bingham visco- damping for force roll-off at low velocities, k0 is the stiffness at large
plastic model, Gamota and Filisko model, Bouc–Wen model, and velocities, k1 is the damper accumulator stiffness, and x0 is the ini-
modified Bouc–Wen model (Spencer, Dyke, Sain, & Carlson, tial displacement of spring. k1 and ab is a third-order polynomial.
1997). Among these techniques, Bingham model is the simplest The corresponding mechanical model is depicted in Fig. 6. In Yang,
352 S. Thenozhi, W. Yu / Annual Reviews in Control 37 (2013) 346–364
Fig. 6. Modified Bouc–Wen model of MR damper.
Fig. 7. Base isolation system.
Spencer, Carlson, and Sain (2002), the dynamic modeling and two
quasi-static models (axisymmetric and parallel-plate model) of
the MR damper are studied through experiments.
The semi-active fluid viscous damper consists of a hydraulic
cylinder, which is separated using a piston head. The cylinder is
filled with a viscous fluid, which can pass through the small ori-
fices. An external valve which connects the two sides of the cylin-
der is used to control the device operation. The semi-active
stiffness control device modifies the system dynamics by changing
the structural stiffness (Symans & Constantinou, 1999).
2.3.4. Base isolators
Base isolators are flexible isolation devices, placed between the
building structure and the foundation for reducing seismic wave
propagation into the structure. The addition of this device will in-
crease the flexibility of the structure, hence the structural time per-
iod. For that reason, isolators reduce the propagation of high
frequency signal from ground to the structure, which makes it suit-
able for implementing in small and middle-rise building structures
(Cheng et al., 2008). Fig. 7 shows the changes in the structure re-
sponse while using base isolator.
Base isolation is one of the popular technique applied widely,
especially in the case of bridges. In general, the isolators can be
formed using elastomeric bearings, sliding bearings, and combina-
tions of both types of bearings. Elastomeric bearings are made up
of elastic materials like the rubber. In the second case, the isolator
uses sliding mechanism (Cheng et al., 2008). In bridges, the isola-
tors are easily implemented by replacing standard bridge bearings
by isolation bearings. More information about the types of isolators
and their implementation can be found in Kelly (2001).
Base isolation is well known passive control technique. But ac-
tive (Chang & Spencer, 2010) and semi-active (Iuliis & Faella, 2013)
control schemes were also proposed. Another class of base isola-
tion devices is the Hybrid Base Isolation (HBI), made by combining
the passive base isolator with the active or semi-active base isola-
tor/control (Spencer & Sain, 1997). Sometimes, the seismic activity
in the building is reduced by placing isolators between the sub-
structure columns, not in the base, hence called seismic isolators.
2.3.5. Hybrid devices
Hybrid actuators combine robustness of the passive device and
high performance of the active devices. Due to the inclusion of
multiple control devices, the hybrid system overcomes the limita-
tions and restrictions seen in the single control devices like passive,
active, and semi-active devices. The hybrid systems are further
classified into two classes: HBI and Hybrid Mass Damper (HMD)
(Spencer & Sain, 1997). As the base isolation exhibits nonlinear
behavior, various nonlinear control technologies like the robust
control were adopted to control these hybrid devices (Forrai
et al., 2001).
HMD can be formed by combining the passive devices like TMD
along with some active devices like AMD. The capability of the
TMD is increased by adding a controlling actuator to it, which in-
creases the system robustness in changing the structure dynamics.
These HMDs are found to be cost effective in terms of the energy
requirement for their operation, when compared with active con-
trol systems (Spencer & Sain, 1997). The full-scale implementation
of active structural control systems in Japan, USA, Taiwan, and Chi-
na are enlisted in Soong and Spencer (2002), where the HMD is
found to be the most commonly employed device compared with
other devices.
Researchers have also investigated the various control methods
for HMD, like optimal control methods, sliding mode control, and
gain scheduling Forrai et al. (2001). As these systems utilize two
types of actuators, it will have a series of objective functions, which
results in a multi-objective optimization problem. To derive an
optimal solution, a preference-based optimization model using
GA is proposed. The designed model is compared with a non-hy-
brid system and is found to be very cost effective in suppressing
the vibrations. A hybrid system using the HMD and a viscous dam-
per is discussed in Park, Ha, Park, and Choo (2009) for the reduc-
tion of the wind induced vibrations of high-rise building.
The implementation of the above mentioned devices will result
in different control schemes, which are summarized in Fig. 8. In the
passive control, the passive device reduces the vibration response
of a structure without using any feedback, see Fig. 8(b). In the ac-
tive and semi-active case, the input and output response of the
structure is measured and based on that the controller generates
a desirable output command signal. This signal is then used to
drive the active or semi-active devices for attenuating the vibra-
tion, which are shown in Fig. 8(c and d), respectively. In the case
of hybrid control shown in Fig. 8(e), only the active/semi-active de-
vice uses the feedback, whereas the passive devices works
independently.
Typical installations of control devices are shown in Fig. 9.
Other recent technique is the connected control method, where
the adjacent buildings are interconnected using control devices
Fig. 8. Control schemes (Soong & Spencer, 2002).
 S. Thenozhi, W. Yu / Annual Reviews in Control 37 (2013) 346–364 353

Consider a passive damper added to a structure represented in

(7), then the system model can be rewritten as Soong and Spencer
(2002)
m€x þ cx_ þ kx þ PðxÞ ¼ ðm þ md Þ€xg ð25Þ
where md is the mass of the damper and P(x) represents the force
corresponding to the damper, used to modify the structure response
for reducing vibrations. The same formulation can be done in the
case of active control devices, where (7) can be rewritten as follows

m€x þ cx_ þ kx ¼ muðtÞ  m€xg ð26Þ

If the control force is selected as per the relationship given in (27)

PðxÞ
uðtÞ ¼ ð27Þ
m
then (26) becomes

m€x þ cx_ þ kx þ PðxÞ ¼ m€xg ð28Þ

Fig. 9. Typical implementation of control devices on structures.
In contrast to the passive control method, here the control func-
tion P(x) is derived as a control law.
The motion equation of a structural system with n-DOF and o
control devices subjected to an earthquake excitation can be ex-
pressed as

M€xðtÞ þ C xðtÞ
_ þ KxðtÞ ¼ CuðtÞ  M K€xg ðtÞ ð29Þ
n1 no
where uðtÞ 2 R is the control force vector and C 2 R is the
location matrix of the control devices. Eq. (29) becomes nonlinear
if the control force is generated using a nonlinear device, such as
MR damper or by using a nonlinear control algorithm, such as intel-
ligent control.

3. Estimation and sensing of structure parameters

Fig. 10. Buildings interconnected using dampers.
for vibration attenuation, see Fig. 10 (Bharti, Dumne, & Shrimali,
2010). In Quiñonero, Massegú, Rossell, and Karimi (2012), passive
devices are installed between the adjacent structures for inter-
structure protection and at the same time semi-active dampers
are placed in the building floors for protecting the substructure.
A brief state-of-the-art review about the structural control de-
vices can be found in Soong and Spencer (2002). The simplicity
of the passive systems made them more common in seismic con-
trol applications. The active systems including the semi-active
and hybrid systems, generates a control force based on the
measurements of the structural responses. Due to this ability of
measuring the structural response it can be designed to accommo-
date a variety of disturbances, which makes them to perform bet-
ter than the passive systems. More on the governing equations of
dampers and actuators can be found in Ahmadizadeh (2007), Ali
and Ramaswamy (2009), Balendra et al. (2001), Cao and Li
(2004), Lin, Dyke, and Rebecca (2007), Mackriell, Kwok, and Samali
(1997), Wang, Ni, Ko, and Spencer (2005), Yalla, Kareem, and
Kantor (2001).
3.1. System identification
In order to identify the parameters of the civil structures, the
dynamic response is studied from its input and output data, and
the parameters are estimated using some sort of identification
techniques. The inputs are the excitation forces like the earthquake
and wind loads, and the outputs are the displacements, velocities,
and accelerations corresponding to the input excitation. In prac-
tice, it is very difficult to derive an exact system model, so the ori-
ginal problem is to obtain parameters, such that the estimated
model responses closely match the output of the building dynamic
behaviors. There exists different methods for identification of both
linear and nonlinear systems (Ljung, 1987).
For the purpose of system identification, the structural system
can be represented using Ordinary Differential Equation (ODE),
transfer functions, state-space models, and Auto Regressive Mov-
ing Average with exogenous input (ARMAX) models (Imai, Yun,
Maruyama, & Shinozuka, 1989). Consider a state-space variable
T
_
zðtÞ ¼ ½xðtÞ; xðtÞ 2 R2n1 , then the system described in (29) can
be represented in state-space form as

z_ ðtÞ ¼ AzðtÞ þ BuðtÞ þ E€xg ðtÞ ð30Þ

2.4. Structure-control device models
yðtÞ ¼ HzðtÞ þ DuðtÞ ð31Þ
Control devices are used to control the dynamics of the struc- where A 2 R2n2n ; B 2 R2nn and E 2 R2n1 .
ture to a desired response. Therefore, the dynamic model of a
structure will change once a control device is installed on it. That  
0 I
is, it is expected that the installation of a control device will modify A¼
M 1 K M1 C
the structure parameters like its natural frequency, thereby chang-
ing the system model (Zhang & Roschke, 1999). As a consequence,    
0 0
it is necessary to consider the dynamics of the actuator in the B¼ ; E¼
M1 C K
structure’s model (Chang, 1999).
354 S. Thenozhi, W. Yu / Annual Reviews in Control 37 (2013) 346–364
Here, the matrices H and D and their dimensions change accord-
ing to the design demands. For example, if the output of the system
is yðtÞ ¼ €
xðtÞ þ K€
xg then
  1
H ¼ M1 K M 1 C ; D ¼ ½M C
System identification can be broadly classified into parametric
and non-parametric identification. In parametric identification,
the system parameters like the mass, stiffness, and damping are
estimated (Zhang & Roschke, 1999). Most commonly used algo-
rithms are least squares method, maximum likelihood method, ex-
tended Kalman filter, and variations of them (Imai et al., 1989).
Non-parametric identification determines a system model from
the measured data, which is a mathematical function that can
approximate the input-output representations sufficiently well
(Mandic & Chambers, 2001). This method is suitable for the sys-
tems with infinite number of parameters. Artificial Neural Network
(ANN) is one of the popular non-parametric identification method
(Hung, Huang, Wen, & Hsu, 2003). Some other known methods are
wavelet networks, splines, and neuro-fuzzy models (Kerschen,
Worden, Vakakis, & Golinva, 2006).
Identification can also be classified into time-domain and fre-
quency-domain, where the identification takes the form of time
series and frequency response functions or spectra, respectively
(Imai et al., 1989; Kerschen et al., 2006). System identification can
be performed either using online or offline techniques. In offline
identification, all the data including the initial states must be avail-
able before starting the identification process. For example, in the
case of building parameter identification, the excitation and the
corresponding structure response are recorded and later used for
identification. Whereas, the online identification is done immedi-
ately after each input-output data is measured. In other words,
the online identification is performed parallel to the experiment,
that is during the structural motion due to seismic or wind loads.
System identification of a linear MDOF structure under ambient
excitation using the eigen space algorithm is presented in Quek,
Wang, and Koh (1999). The algorithm identifies the damping and
stiffness with known mass. In Xu, Wu, Chen, and Yokoyama
(2004), two Back-Propagation Neural Networks (BPNN) are used
to estimate the stiffness and damping of a 5-story building, where
the first one is called emulator NN and the second one is known as
the parametric evaluation NN. A modified GA strategy (Perry, Koh,
& Choo, 2006) and GA with gradient search (Zhang, Koh, & Duan,
2010, 2010) is proposed to improve the accuracy and computa-
tional time for parameter identification of MDOF structural sys-
tems. Sometime, the parameters are identified in the structure
equipped with the actuator (Dyke, Spencer, Quast, Kaspari, & Sain,
1996; Fabián & Icaza, 2010). On the other hand, identification is
performed only for the control devices. In Boada, Calvo, Boada,
and Díaz (2011), a memory based learning called lazy recursive
learning method based on NN is used to identify the MR damper
behavior. The input current to the MR damper is varied and the
corresponding damper behavior is modeled.
System identification is sometimes used for modal analysis,
where the modal parameters like natural frequencies (xn) for differ-
ent modes, modal shapes, and damping ratios (f) of the structures
are estimated (Kerschen et al., 2006). One such a simple technique
is the analysis using Fourier transform techniques to estimate
power spectra from which the modal parameters are estimated
(Jones, Shi, Ellis, & Scanlan, 1995). When the input excitation fre-
quency equals the structure natural frequency, the magnitude of
the vibration becomes higher. So it is important to estimate these
low order natural frequencies and to control the structure from
vibrating at those frequencies. A modified random decrement meth-
od along with Ibrahim time-domain technique is used for estimat-
ing the modal parameters, which uses the floor acceleration (Lin,
Hong, Ueng, Wu, & Wang, 2005). The modal parameters can also
be identified using Kalman filter (Adhikari, Yamaguchi, & Yamazaki,
1998; Allen, Zazzera, & Scattolini, 2000; Choi et al., 2005).
Parametric identification of a linear structure excited with two
orthogonal horizontal components using least-squares identifica-
tion algorithm is presented in Cervantes and Icaza (2010). Here
each floor is considered to have 3-DOF, two displacements (along
the x and y axis) and one torsion (rotation around the z axis). In
Hong, Betti, and Lin (2009), the dynamic state-space model of an
earthquake-excited structure is identified using the measured in-
put-output data that is used later for estimating the modal param-
eters. The system and modal parameters of a linear MDOF structure
is estimated in Lee and Yun (1991). Here the equation of motion of
the structure is first written in state-space equation of the observa-
ble canonical form and then is converted into an ARMAX model for
dealing with the noise present in the measured data.
Some works (Ali & Ramaswamy, 2009; Jiang et al., 2010; Shook,
Roschke, Lin, & Loh, 2008) consider the damping matrix C as a Ray-
leigh damping coefficient matrix, which is found using the modal
parameters as given below (Chopra, 2001),
C ¼ aR M þ bR K ð32Þ
where the Rayleigh parameters aR and bR are calculated using the
first and third eigen-frequencies (x1 and x3), given by
2fx1 x3 2f
aR ¼ and bR ¼ ð33Þ
x1 þ x3 x1 þ x3
whereas (Jiang et al., 2010) uses the first two lower-order mode
frequencies.
In Shook et al. (2008), Zhang (2010), the stiffness of the struc-
ture column is estimated using the equation given below
12Ey Im
k¼ ð34Þ
L3c
where Ey is the Young’s modulus of elasticity, Im is the moment of
inertia, and Lc is the unsupported length of the column.
A brief review about the identification of nonlinear dynamic
structures is presented by Kerschen et al. (2006) in 2006. The fun-
damentals and methods of identification for linear and nonlinear
structural dynamic systems are reviewed in Imai et al. (1989). A
general survey on system identification is presented in Åström
and Eykhoff (1971) and a review on stochastic identification meth-
ods for modal analysis is presented in Peeters and Roeck (2001).
3.2. Sensing and estimation of system states
In order to control the structural dynamics, it is necessary to
measure the system states directly using a sensor or indirectly
by using a state observer. Some structural control applications
use Kalman filter as the observer for estimating the velocity and
displacement (Gu & Oyadiji, 2008; Park, Kohb, Okb, & Seo, 2005;
Xu & Guo, 2008; Yi & Dyke, 2000). A Kalman filter estimator is gi-
ven by
^z_ ¼ A^z þ Bu þ Lðy  H^z  DuÞ ð35Þ
1 T T
L ¼ R ðcg FE þ HSÞ ð36Þ
where ^z is the estimate of the state vector z, L is the Kalman filter
gain matrix, S is the solution of the Algebraic Riccati equation using
matrix R, and cg is the power spectral density of ground acceleration
to the sensor noise. In Zhang and Roschke (1999), the Kalman-Bucy
filter is used as the state estimator represented by
^z_ ¼ A^z þ Bu þ Lðy  y
^Þ ð37Þ
T 1
L ¼ EC R ð38Þ
 S. Thenozhi, W. Yu / Annual Reviews in Control 37 (2013) 346–364
Kalman filter cannot be applicable if the building parameters;
mass, stiffness, and damping are not available, in that case sensors
are used for the state estimation. There are different sensors avail-
able to measure displacement, velocity, and acceleration (Housner
et al., 1997). During the seismic excitation, the reference where the
displacement and velocity sensors are attached will also move, as a
result the absolute value of the above parameters cannot be
sensed. Alternatively, accelerometers can provide inexpensive
and reliable measurement of the acceleration at strategic points
on the structure. A comparative study about the performance of
the displacement, velocity, and acceleration sensors are performed
in Balendra et al. (2001), Xu (1996) and it is shown that the accel-
eration sensor is more effective compared to the other two sensors.
A number of experiments and implementations about the acceler-
ation feedback in structural control were carried out in Chung, Wu,
and Jin (1998), Dyke et al. (1996, 1996).
An accelerometer measures the absolute acceleration, which is
then integrated for estimating the velocity and displacement.
Obtaining the velocity and displacement from the measured accel-
eration is a practically challenging task. Although time integration
of the acceleration seems to be a straightforward solution for esti-
mating the velocity and displacement, there are some practical dif-
ficulties that can result in a wrong estimation. Integrating these
signals will result in the amplification of low frequencies compo-
nents, reduction in the magnitude of high frequencies signals,
and phase errors. In other words, the low frequency signals includ-
ing the DC offset present in the acceleration signal will dominate
the result of the velocity and displacement, giving an unrealistic
estimation.
The output of the accelerometer a(t) can be expressed as
aðtÞ ¼ ka €xðtÞ þ uðtÞ þ e ð39Þ
where ka is the accelerometer gain, u(t) is the noise and disturbance
effects of the measurement, and e denotes the DC bias (Link & von
Martens, 2004; Zhu, 2007). Accelerometer has different source of
noise, integrating these noise signals leads to an output that has a
Root Mean Square (RMS) value that increases with integration time,
even in the absence of any motion of the accelerometer (Thong,
Woolfson, Crowe, Gill, & Jones, 2004). The RMS positional error
ex(t) of an acceleration signal with a bias e can be approximated as
1 2
RMSfexðtÞ g ¼ et ð40Þ
2
which will grow at a rate of t2.
It has been shown that the aliasing can cause low-frequency er-
rors in the measured acceleration signal (Edwards, 2007). Aliasing
is an unavoidable phenomenon, that happens when digitizing the
analog signals using an Analog-to-Digital Converter (ADC). During
this conversion, the frequency components above the Nyquist rate
are folded back into the bandwidth of interest. Then, the accelera-
tion signal in (39) can be rewritten as
aðtÞ ¼ ka €xðtÞ þ uðtÞ þ e þ €xs ðtÞ ð41Þ
where €xs ðtÞ is the aliasing content due to sampling. This low fre-
quency content will be amplified during the integration process.
This aliasing effect is not completely removable but its effect can
be minimized by using an anti-aliasing filter between the acceler-
ometer and data acquisition card. The ADC sampling rate needs to
be high enough compared to this filter cutoff frequency and the
sampling should to be done in uniform time intervals.
The other source of offset in the measured acceleration is the
ADC itself (Boore, 2003). If the acceleration is slow compared with
the quantization level of the conversion, an offset is added into the
acceleration signal. This effect can be reduced by increasing the
resolution of the ADC.
355
Apart from these issues, the integration output can also be af-
fected by the integration techniques. The integration methods like
the Trapezium rule, Simpson’s rule, and Tick’s rule have problems
with low-frequency components, and they also show instability at
high frequencies (Worden, 1990).
A drift-free integrator is proposed by Gavin, Morales, and Reilly
(1998) , which is implemented using analog and digital circuits.
The paper presents three types of integrators: (1) implemented
using a first order low-pass filter as the integrator and two stages
of high-pass-filters for removing the offset, (2) analog integrator
with feedback stabilization, and (3) a stabilized hybrid analog-dig-
ital integrator with an exponential accuracy when integrating
long-period signals. In another work (Ribeiro, de Castro, & Freire,
2003), the drift due to the integration is eliminated by; first filter-
ing the acceleration signal using a frequency domain filter called
Fast Fourier Transform-Direct Digital Integration (FFT-DDI) and
then is integrated for estimating the velocity and displacement.
The same method is repeated for removing the drift occurred due
to the unknown initial conditions.
The constant offset present in the acceleration data can be rep-
resented using a baseline. The integration may cause a drift in this
baseline, which will give a wrong estimation. A baseline correction
method is proposed in Yang, Li, and Lin (2006) that uses a least-
square curve fitting technique and a frequency domain filtering
for avoiding the drift during the integration. The correction is done
by determining a baseline in polynomial form, which is then sub-
tracted from the measured acceleration signal, then is integrated to
obtain the velocity and displacement. Finally, a windowed filter is
applied to remove the low-frequency noise.
A practical method for calibrating the positional error obtained
by double integrating the acceleration signal is discussed in Thong
et al. (2004). The double integration of noise using different tech-
niques is also presented. An initial velocity determination method
for the displacement estimation from the acceleration data is sug-
gested in Park, Kim, Park, and Lee (2005), which also considers the
initial condition in their design. A weighted residual parabolic
acceleration time integration method is proposed in Razavi,
Abolmaali, and Ghassemieh (2007), where the displacement is as-
sumed to be a fourth order polynomial, so that the acceleration
variation with time is quadratic. A numerical integrator for esti-
mating the velocity and displacement from the measured acceler-
ation signal is proposed in Thenozhi, Yu, and Garrido (2013). The
effectiveness of the integrator is illustrated experimentally by per-
forming a structural vibration control on a shake table using a PD
controller.
4. Control of building structures
The objective of structural control system is to reduce the vibra-
tion and to enhance the lateral integrity of the building due to
earthquakes or large winds, through an external control force
(Kim & Yun, 2000; Marzbanrad, Ahmadi, & Jha, 2004; Zhang,
2010). In active control system, it is essential to design one control-
ler in order to send an appropriate control signal to the control de-
vices so that it can reduce the structural vibration. The control
strategy should be simple, robust, fault tolerant, need not be an
optimal, and of course must be realizable (Tang, 1996).
4.1. Time-delay
One of the main challenges in structural control system is the
time-delay, which may occur in different stages of the systems like
in data acquisition, data processing, sophisticated control algo-
rithms, control device, or the sum of these effects (Du & Zhang,
2008). Among these delays, the delay in control force caused by
356 S. Thenozhi, W. Yu / Annual Reviews in Control 37 (2013) 346–364
large mechanical control devices will affect the properties of the
building structure models. The inclusion of time-delay in the con-
troller design provides a more realistic model for the structural
vibration control applications.
These delays may cause instability in closed-loop systems
(Agrawal, Fujino, & Bhartia, 1993; Agrawal & Yang, 1997; Datta,
2003; Symans & Constantinou, 1996; Xue, Sun, Wu, & Zhang,
2011; Zhang & Roschke, 1999). A state-of-the-art review on the
fixed time-delay effects in actively controlled civil engineering
structures is presented in Agrawal and Yang (1997), which also dis-
cuss the effect of time-delays on the stability and performance of
the system. The controlled system will become unstable, if the
time-delay is greater than a delay known as critical time-delay.
The equation of motion of a SDOF structure with a control de-
vice, in the frequency domain is given by Agrawal and Yang (1997)
€x þ 2fxn x_ þ x2n x ¼ uðtÞ þ f ðtÞ ð42Þ
_
where uðtÞ ¼ g 1 xðtÞ  g 2 xðtÞ is the control force, where g1 and g2
are the displacement and velocity feedback gains, respectively.
If a fixed time-delay td is present in the control force, it can be
represented in the Laplace form as
F d ðsÞ ¼ estd ðg 1 s þ g 2 Þ ð43Þ
Finally, the closed-loop transfer function of the above system can be
represented as
PðsÞ
TðsÞ ¼ ð44Þ
1 þ PðsÞF d ðsÞ
1
where PðsÞ ¼ ðs2 þ 2fxn s þ x2n Þ .
The phase margin of the open-loop system P(s)Fd(s) will be re-
duced by the delay term estd . Then, the critical time-delay tdmax
is defined as the time-delay at which the open-loop transfer func-
tion becomes zero, which makes the system unstable when td > -
tdmax. It is also concluded that, due to the presence of the delay
term estd , the characteristic equation of the above system will
have an infinite number of roots. So in order to ensure the system
stability, all the roots should have negative real parts.
The equation of motion of n-DOF structure with time-delay in
the control force can be represented as
M€xðtÞ þ C xðtÞ
_ þ KxðtÞ ¼ Cuðt  td Þ  M K€xg ðtÞ ð45Þ
Then, (30) can be rewritten as
z_ ðtÞ ¼ AzðtÞ þ Buðt  t d Þ þ E€xg ðtÞ ð46Þ
The stability analysis method and critical time-delay calculation
for a n-DOF system under single and multiple actuator cases were
also presented in Agrawal and Yang (1997). A review on time-delay
compensation methods is presented in Agrawal and Yang (2000).
4.2. Sensor and actuator placement
The optimal placement is concerned with placement of the
sensing and controlling devices in preselected regions in order to
closely perform the measurement and control operation of the
structure vibration optimally. The actuator and sensor play an
important role in deciding the system’s controllability and observ-
ability, respectively. So it is important to perform an optimal place-
ment of the sensors and actuators such that the controllability and
observability properties of all or selected modes are maximized.
Due to the above mentioned reasons and importance, a number
of studies are carried out about the optimal placements of devices
(Amini & Tavassoli, 2005; Flynn & Todd, 2010; Gawronski, 1997;
Guney & Eskinat, 2008; Hiramoto, Doki, & Obinata, 2000; Li, Tang,
& Li, 2004; Liu, Yang, & Li, 2003; Liu, Hou, & Demetriou, 2006; Obe,
1985). A survey on the optimal placement of control devices can be
found in Fisco and Adeli (2011).
In Gawronski (1997), the actuator and sensor location perfor-
mance index is calculated between the c2WZ and c2UY Hankel singular
values. A non-negative correlation coefficient j is defined as
 
c2 T c2UY
j2 ¼ 2 WZ
ð47Þ
cWZ 2 c2UY 2
where c2WZ and c2UY represents the Hankel singular values of the
transfer functions GWZ and GUY, respectively. Here U and W are
the inputs to the system and Y and Z are the outputs of the system.
As per the above equation, the maximal performance is obtained
with a better controllability and observability properties when j
reaches a maximal value; j = 1, which is achieved when c2UY ¼ c2WZ .
A closed-loop optimal location selection method for actuators
and sensors in flexible structures is developed by Guney and Esk-
inat (2008), which uses a simple H1 controller where the location
optimization is performed using a gradient-based unconstrained
minimization. Another related work is done in Liu et al. (2006)
using a H2 norm based computation for a reduced model of flexible
structures, which considers only the dominant modes. They also
proposed one GA for the nonlinear optimization problem for the
reduced order model. A GA is proposed in Liu et al. (2003) through
the formulation of a discrete and nonlinear optimization problem.
Finally, the proposed algorithm is simulated for a 16-story building
under 18 different earthquake excitations. In the work (Li et al.,
2000), it is concluded that the optimal position of actuators de-
pends on the control algorithm, so that different control algorithms
or different controllers yield different positions of the actuators.
4.3. Linear control of building structures
4.3.1. PID control
The Proportional-Integral-Derivative (PID) has been widely con-
ducted for practical applications, especially for the systems with
one or two DOF. For multivariable systems, its control algorithm
becomes more complex, which makes them unsuitable for the
applications like vibration control of MDOF flexible structures. A
simulation was carried out for a simple proportional controller,
which is able to reduce the building displacement for wind excita-
tion, but found to be ineffective for strong earthquake excitation
(Nerves & Krishnan, 1995).
In Guclu and Yazici (2008), two PD controllers were used for
controlling two actuators installed in the first and fifteenth floor.
The control law is given as
 
deðtÞ
uðtÞ ¼ K p eðtÞ þ K d ð48Þ
dt
where Kp and Kd are the proportionality constant and derivative
time, respectively and e(t) is the position error. The designed PD
controller performance is found to be less efficient when compared
with that of a Fuzzy Logic Controller (FLC).
In a work done by Guclu (2006), a PID controller is designed
which have the following controlling law
 
Z t
1 deðtÞ
uðtÞ ¼ K p eðtÞ þ eðtÞdt þ K d ð49Þ
Ki 0 dt
where Ki is the integral gain. Here the PID performance is compared
with that of a Sliding Mode Controller (SMC) and found to be less
effective in controlling the structural vibration. In Tinkir, Kalyoncu,
and Sß ahin (2013), a Proportional-Integral (PI) controller is used to
actuate the AMD against the structural motion due to earthquake.
 S. Thenozhi, W. Yu / Annual Reviews in Control 37 (2013) 346–364
4.3.2. H1 control
H1 technique is one of the widely used linear robust control
scheme in structural vibration control. This technique is insensitive
with respect to the disturbances and parametric variations, which
makes them suitable for the MIMO type structural control systems
(Utkin, 1990).
A modified H1 controllers, for example, pole-placement H1
control is presented in Park et al. (2008). In this work, instead of
changing the structure stiffness some target damping ratio is con-
sidered. A bilinear transform is adopted to locate the closed-loop
poles in a specific region within the H1 controller design frame-
work. The relation between the final closed-loop poles and bilinear
transform parameters is derived as a quadratic equation and using
this equation a new non-iterative direct method is developed for
an optimal H1 controller design.
Normally, the H1 design results in a higher order system, which
will make the implementation more difficult (Saragih, 2010). So it
may be necessary to reduce its order, which can be done by per-
forming balanced truncation. The truncation has two classes; di-
rect method and indirect method. The balanced truncation
assures very few information losses about the system, which is
achieved by truncating only less controllable and observable
states. It is shown that the performance of the reduced low order
system is nearly same as the performance of the actual higher or-
der controller.
A H1 based structural controller using Takagi-Sugeno Fuzzy
model was proposed in Chen, Chiang, Hsiao, and Tsai (2004,
2009). The controller stability is derived based on Lyapunov stabil-
ity theory, which is evaluated as a LMI problem. If the initial con-
dition is considered, the H1 control performance satisfies the
following condition:
Z tf Z tf
zT ðtÞQzðtÞdt 6 zT ð0ÞPzð0Þ þ 2 €xTg €xg dt
0 0
where tf denotes the termination time of the control, P and Q are
the positive definite matrices, and  denotes the effect of €xg on
z(t). The effectiveness of the proposed algorithm is demonstrated
through numerical simulations on a 4-story building.
As discussed earlier, time-delay is an important factor to be
considered while designing a control system. A H1 controller is
presented in Du and Zhang (2008), which considers time-delay in
control input u as referred in (45). The proposed algorithm deter-
mines the feedback control gain with a random search capability
of GA and solving a set of LMIs. The effectiveness of the proposed
algorithm is proved through simulation of a system with larger in-
put time-delay.
4.3.3. Optimal control
Optimal control algorithms are based on the minimization of a
quadratic performance index termed as cost function, while main-
taining a desired system state and minimizing the control effort
(Nerves & Krishnan, 1995). The most basic and commonly used
optimal controller is the Linear Quadratic Regulator (LQR). For
structural control applications, the acceptable range of structure
displacement and acceleration are considered as the cost function
that is to be minimized.
An energy based LQR is proposed in Alavinasab and Moharrami
(2006), where the controller gain matrix is obtained by considering
the energy of the structure. The structural energy is defined as
1 T 1
_ þ xT ðtÞKxðtÞ
x_ ðtÞMxðtÞ
2 2
where the first term is the kinetic energy and the second term is the
potential energy of the structural system.
A modified LQR is proposed in Djajakesukma et al. (2002),
which is formed by adding an integral and a feed-forward control
357
to the classic LQR. A state feedback gain and an integral gain are
used to reduce the steady-state error. A feed-forward control is in-
cluded to suppress the structural responses and to reduce the ef-
fect of earthquakes. A structural vibration control utilizing a
filtered LQ control is presented in Seto (1996). As all the structural
state variables are not observable, a sub-optimal control is used,
where the system states are reduced using low-pass filters. A
LQR based on GA is presented (Jiang et al., 2010), where the GA
is used for choosing the weighting matrix.
Sometime, states of the structures are measured indirectly
using some observers like Kalman filters. The addition of a Kalman
filter to a LQR control strategy leads to what is termed as Linear
Quadratic Gaussian (LQG) (Zhang & Roschke, 1999). In other words,
LQG is formed by combining the linear quadratic estimator with
LQR. These LQG are generally used for the systems which has
Gaussian white noise (Fisco & Adeli, 2011).
The conventional LQG controller sometimes do not consider the
input force term in their design. Ho and Ma (2007) proposed an ac-
tive vibration control scheme using a combination of LQG and an
input estimation approach. The input estimation approach is intro-
duced to observe the input disturbance forces for the open loop
control, that is used to cancel out the input forces. The proposed
method is evaluated through numerical experiments on linear
lumped-mass systems and a better performance is reported com-
pared to that of the conventional LQG.
An active controller utilizing MR damper is designed using LQG
control strategy under a wind loading by means of drag forces
(Zhang & Roschke, 1999). A real set of recorded wind speed data
is used to excite the laboratory prototype. A H2/LQG based control-
ler is presented in Lin et al. (2007), which uses wireless sensing
motes (MICA2) for sensing the acceleration signal. More works
about the structural control using LQR/LQG control algorithms
ð50Þ can be found in Pourzeynali, Lavasani, and Modarayi (2007), Wang,
Roschke, and Yeh (2002), Yi and Dyke (2000). Optimal algorithms
based on instantaneous optimal control has also been developed
for nonlinear systems. The nonlinear optimal methods using GA,
FLC, etc., will be discussed later.
4.4. Nonlinear control of building structures
4.4.1. Sliding mode control
SMC is one of the most popular robust control techniques. A
switching control law is used to drive the system’s state trajectory
onto a pre-specified surface in the state-space and to maintain the
system’s state trajectory on this surface for subsequent time, which
results in a globally asymptotically stable system. In the case of
structural vibration control, this surface corresponds to a desired
system dynamics. The robustness of the SMC against the uncer-
tainties and parameter variations makes them a better choice for
structural control applications.
The nonlinear control force in SMC is given as
u ¼ ueq  g sgnðrðtÞÞ ð52Þ
where the linear term ueq is the equivalent control force, r = [r1, -
. . . , rn] are the n sliding variables, and g is the design parameter that
guarantees the system trajectories reach the sliding surface in finite
time. A SMC with hybrid control is proposed in Zhao, Lu, Wu, and
Mei (2000), where the control law also termed as reaching law is
formed using the constant plus proportional rate reaching law
and power rate reaching law.
ð51Þ Due to the imperfection in the high-frequency discontinuous
switching, the direct implementation of the control given in (52)
will result in chattering effect, which may cause damage to the
mechanical components, hence the actuators. This effect should
be eliminated by suitably smoothing the control force or by using
continuous SMC. Many structural control strategies based on the
358 S. Thenozhi, W. Yu / Annual Reviews in Control 37 (2013) 346–364
non-chattering SMC were reported (Adhikari & Yamaguchi, 1997;
Allen et al., 2000; Guclu, 2006; Yagiz, 2001; Yang, Wu, Agrawal,
& Hsu, 1997; Zhao et al., 2000).
A Modal Space Sliding-Mode Control (MS-SMC) method is de-
signed in Adhikari et al. (1998), where the dominant frequencies
are derived using power spectrum as well as the wavelet analysis
of the time series of the input-output. SMC based on a single-mode
(first mode) reduced-order model is designed. Another SMC based
on the modal analysis is presented in Allen et al. (2000), where the
first six modes of the structure were considered.
During seismic events, the main control unit may lose its func-
tionality, so it is a better option to use a decentralized system,
where the whole control is divided into subsystems and are con-
trolled independently. Such a type of decentralized system with
SMC is presented in Nezhad and Rofooei (2007). The numerical
studies were carried out for full control and partial control cases
and reaching laws were derived for cases; with and without con-
sidering actuator saturations. They found that the full control case
is more effective, and they could not find any significant changes in
the control for different subsystem configurations.
A NN based SMC for the active control of seismicity excited
building structures is proposed in Yakut and Alli (2011). Here apart
from the sliding variables, the matrix r also represents the slope of
the sliding surface. This slope moves in a stable region, which re-
sults in a moving sliding surface. A four layer feed-forward NN is
used to reduce chattering effect and to determine the sliding sur-
face slope. To achieve a minimum performance index, the control-
ler is optimized using a GA during the training process. It is shown
that a high performing controller is achieved by using the moving
sliding surface. Another SMC based on Radial Basis Function (RBF)
NN is reported in Li, Deng, and Gu (2010). The chattering free SMC
is obtained using a two-layered RBF-NN. The relative displacement
of each floor is fed as the input to the NN and the design parameter
g is taken as the output. A modified gradient-descent method is
used for updating the weights.
Couple of research works were carried out in designing the SMC
using Fuzzy logic so called, Fuzzy Sliding Mode Control (FSMC) (Alli
& Yakut, 2005; Kharaajoo, Kangharloo, & Roudsari, 2004; Kim &
Yun, 2000; Wang & Lee, 2002; Wang & Lin, 2007). The SMC pro-
vides a stable and fast system, whereas the FLC provides the ability
to handle a nonlinear system. The Chattering problem is avoided in
most of these FSMC systems. A FSMC based on GA is presented in
Wang et al. (2002), where the GA is used to find the optimal rules
and membership functions for the FLC.
4.5. Intelligent control of building structures
4.5.1. Neural network control
In recent years, the structural control systems based on NN are
very popular, because of its massively parallel nature, ability to
learn, and its potential in providing solutions to the foregoing un-
solved problems. They provide a general framework for modeling
and control of nonlinear systems such as building structures.
In the middle of 1990s, very few structural control applications
have been reported based on NN. Wen, Ghaboussi, Venini, and Nik-
zad (1995) presented a NN based active control of a SDOF system
that can become nonlinear and inelastic. One inverse mapping
NN and one emulator NN are used in the design. The difference be-
tween the actual overall structural response and response due to
the control force only, is used as the input to the inverse mapping
NN. The emulator NN predicts the response of the structural sys-
tem to the applied control force. Using this response, a control
force with a phase-shift is generated to nullify the excitation.
A Back-Propagation (BP) based ANN for active control of SDOF
structure is proposed by Tang (1996). This control strategy does
not need the information of the external excitation in advance
and the control force needed for the next sampling time is com-
pletely determined from the currently available information. The
ANN with five neuron elements (displacement, velocity and load
of the preceding time step and displacement and velocity of the
current time step) is used, which will perform two sequential cal-
culations in every sampling interval; (a) calculate the load (b)
based on the calculated load, the control force u(t) needed for the
next time interval is calculated. Apart from the numerical verifica-
tion of the above algorithm, they have also presented a study on
the uncertainties in the system modeling and input motion.
Consider the minimization of the cost function in a discrete
form with total time step r and increment time Dt
X r X r
bJ ¼ bJ n^ ¼ 1 ^ T Qz½n
ðz½n ^ T Ru½n
^  þ u½n ^ÞDt ð53Þ
^ ¼0
n
2 ^ ¼0
n
where bJ n^ is the instantaneous cost function, bJ is the global cost
function, and n ^ is the discrete time steps. If the weights are up-
dated at each time step in order to minimize the instantaneous
cost function, this learning mode is called pattern learning, and
if the weights are updated once for all time steps so that the glo-
bal cost function bJ is reduced, this learning mode is known as
batch learning. An optimal control algorithm using NN based on
the pattern learning mode is presented in Kim, Jung, and Lee
(2000). The steepest-descent method is used here as the weight
updating rule.
One multi-layer NN controller with a single hidden layer is pre-
sented in Cho, Fadali, Saiidi, and Lee (2005). The optimal number of
hidden neurons is selected after performing a number of iterative
training cycles. The network will generate an active control force
as output using the structure response as its input. The batch learn-
ing is used here, where the network weights and biases are se-
lected in such a way that a minimal objective cost function is
achieved. The steepest-gradient-descent optimization method is
used for the weight update, where the partial-differential equa-
tions are solved using the chain rule.
Probabilistic Neural Networks (PNN) are feed-forward networks
built with three layers. They are derived from Bayes decision net-
works that estimates the probability density function for each class
based on the training samples. The PNN trains immediately but
execution time is slow and it requires a large amount of memory
space. A new method to prepare the training pattern and to calcu-
late PNN output (control force) quickly is proposed in Kim, Kimb,
Chang, and Jung (2008). The training patterns are uniformly dis-
tributed at the lattice point in state-space, so that the position of
invoked input can be known. This type of network is called as Lat-
tice Probabilistic Neural Network (LPNN). The calculation time is
reduced by considering only the adjacent patterns. Here the dis-
tance between the input pattern (response of structure) and train-
ing patterns (lattice type) for LPNN are calculated, which is then
converted as the weights.
An active type NN controller using one Counter-Propagation
Network (CPN) is presented in Madan (2005), which is an unsuper-
vised learning type NN, so that the control force is generated with-
out any target control forces. Another intelligent control technique
using a NN is proposed for seismic protection of offshore structures
(Kim, 2009).
The ability of the nets to perform nonlinear mappings between
the inputs and outputs, and to adapt its parameters so as to mini-
mize an error criterion, make the use of ANN particularly well sui-
ted for the identification of both linear and nonlinear dynamic
systems. The NN for system identification in structural control
applications were presented in Chassiakos and Masri (1996), Chen
(2009), Liu, Xia, and Zhu (2009), Tani, Kawamura, and Ryu (1998),
Xu et al. (2004). A NN is designed to approximate the nonlinear
structural system and the corresponding stability conditions are
 S. Thenozhi, W. Yu / Annual Reviews in Control 37 (2013) 346–364
derived (Chen, 2009). A state-feedback controller for the NN is de-
signed using a Linear Differential Inclusion (LDI) state-space repre-
sentation, which is useful in the stability analysis. Using NN, the
system in (30) is approximated as a LDI representation with less
modeling errors.
An intelligent structural control system with improved BP–NN
is proposed in Liu et al. (2009), which is used to predict the inverse
model of the MR damper and for eliminating time-delay in the sys-
tem. The system represented in (46) is considered here. The system
has two controllers; the first one modifies the actual structural
model, which was offline trained before and the second controller
causes error emendation by means of online feedback. A multi-
layer NN for structural identification and prediction of the earth-
quake input is presented in Tani et al. (1998).
4.5.2. Fuzzy logic control
Like NN, Fuzzy logic is also a model free approach for system
identification and control. The FLC design involves; the selection
of the input, output variables, and data manipulation method,
membership function, and rule base design. Due to its simplicity,
nonlinear mapping capability, and robustness, the FLC has been
used in many structural control applications (Aldawod, Naghdy,
Samali, & Kwok, 1999; Ali & Ramaswamy, 2009; Choi et al.,
2005; Guclu & Yazici, 2008; Park et al., 2002; Park et al., 2004; Reit-
hmeier & Leitmann, 2001; Shook et al., 2008; Teng, Peng, & Chuang,
2000; Yalla et al., 2001; Yan & Zhou, 2006; Yeh, Chiang, & Juang,
1994; Zhang, 2010).
A FLC is designed (Guclu & Yazici, 2008) for a 15-story struc-
ture with two type of actuators, one mounted on the first floor
and the other actuator (ATMD) on the fifteenth floor. The pro-
posed FLC uses the position error and their derivatives as the in-
put variable to produce the control forces for each actuator. The
rule base is formed using seven fuzzy variables. The controller
uses Mamdani method for fuzzification and Centroid method for
defuzzification. A simulation using Kocaeli earthquake signal is
carried out to prove the improvement in the performance of the
FLC. A similar type of FLC is presented in Aldawod et al. (1999),
for the active control of wind excited tall buildings using ATMD.
Another FLC for MDOF is proposed (Yeh et al., 1994), that uses
the same architecture, which is further modified into MDOF using
weighted displacement and weighted velocity. In order to get the
maximum displacement and velocity values, a high magnitude
earthquake is used to excite the building structure. As all the
floors do not have control devices, a weighting value is assigned
to each floor, which will be large if the control device is closer
to that particular floor. Finally, a force factor is calculated based
on the weights of each floor.
A Fuzzy based on-off controller is designed to control the struc-
tural vibration using a semi-active TLCD (Yalla et al., 2001). The
optimal control force is given as
X
r
u¼ pi zi ð54Þ
i¼1
where pi = [p1, . . . , pr] is the optimal control gain vector obtained
using LQR strategy. The control force will act opposite to the direc-
tion of the liquid velocity ðx_ f Þ. The regulation of the control force is
done by varying the coefficient of headloss (n) with the semi-active
control rule as given below.
nmax if fzl ðtÞx_ f ðtÞg < 0
nðtÞ ¼ ð55Þ
nmin if fzl ðtÞx_ f ðtÞg P 0
where zl represents the largest weighted state, which contributes
most of the control force in (54). Finally, using the above control
law a FLC is designed, that takes the liquid velocity and the large
weighted displacement (zi = zl) as its input and produces the
359
coefficient of headloss as the output, which is used to control the
valve in the semi-active TLCD.
A fuzzy supervisory control method is presented in Park et al.
(2002), which has a fuzzy supervisor in the higher level and three
sub-controllers in the lower level. First, the sub-controllers are de-
signed based on the LQR strategy, where the three sub-controllers
are derived from three different weight matrices. The fuzzy-super-
visor tunes these sub-controllers according to the structure’s cur-
rent behavior. A similar work is done in Park et al. (2004), where
the sub-controller is designed using an optimal controller in the
modal space. The matrix in the Riccati equation is calculated using
the natural frequencies of the dominant modes and a correspond-
ing gain matrix is determined. Another FLC for active control of
structure using modal space is presented in Choi et al. (2005),
which uses a Kalman filter as an observer for the modal state
estimation and a low-pass filter for eliminating the spillover
problem.
Instead of using a mathematical model, a black-box based
controller is proposed in Das, Datta, and Madan (2012). Here the
force-velocity characteristics of the MR damper corresponding to
different voltages are obtained experimentally, which are used to
calculate the desired control force. The effect of the damper posi-
tion and capacity on the control response is also studied.
An alternative to the conventional FLC, using an algebraic meth-
od is proposed in Duc, Vu, Tran, and Bui (2011). Here the hedge
algebra is used to model the linguistic domains and variables and
their semantic structure is obtained. Instead of performing fuzzifi-
cation and defuzzification, more simple methods are adopted,
termed as semantization and desemantization, respectively. The
hedge algebra based fuzzy system is a new topic, which was first
applied to fuzzy control in 2008. Compared to the classic FLC, this
method is simple, effective, and can be easily interpreted.
Some structural vibration controllers were designed, where the
FLC is combined with the GA (Ali & Ramaswamy, 2009; Pourzey-
nali et al., 2007; Shook et al., 2008; Yan & Zhou, 2006). The GA is
known for its optimization capabilities. The GA is used here to opti-
mize different parameters in the FLC like its rule base and member-
ship function.
4.5.3. Genetic algorithm
The GA is an iterative and stochastic process that proceeds by
creating successive generation of offsprings from parents by per-
forming the operations like selection, crossover, and mutation.
The above operation is performed based on the fitness (termed
as cost function in optimization problems) value assigned to each
individual. After these operations, the parents are replaced by the
offsprings, which is continued till an optimal solution for the prob-
lem is attained (Fleming & Purshouse, 2002).
The structural control problem consists of different objectives
to be optimized, which can be formulated using multi-objective
optimization algorithms like GA. In Park and Koh (2004), a prefer-
ence-based optimum design using GA for an active control of struc-
ture is proposed, where the structure and control system is treated
as a combined system. Here the structural sizing variables, loca-
tions of actuators, and the elements of the feedback gain matrix
are considered as the design variables and the cost of structural
members, required control efforts, and dynamic responses due to
earthquakes are considered as the objective functions to be mini-
mized. For each objective criterion, preference functions are de-
fined in terms of degrees of desirability and regions that
represent the degrees of desirability. They are categorized as desir-
able, acceptable, undesirable, and unacceptable with ranges de-
fined by ðki 6 ci1 Þ; ðci1 6 ki 6 ci2 Þ; ðci2 6 ki 6 ci3 Þ, and ðci3 6 ki Þ
respectively, where ci1 ; ci2 , and ci3 are the range boundary values
and ki is the ith design objective. The preference-based optimiza-
tion problem model is then given as
360 S. Thenozhi, W. Yu / Annual Reviews in Control 37 (2013) 346–364
1X l
F P ðkÞ ¼ fP ½ki ðdÞ
l i¼1 i
with ki ðdÞ < ci3 and dmin 6 d 6 dmax, where FP is the aggregate pref-
erence function, fP is the power function, l is the number of design
objectives, and d is the vector of design variables; dmin and dmax are
the prescribed design constraints, respectively. Finally, the fitness
function of ng randomly created strings is defined as follows
F fi ¼ ½maxðF Pj Þ þ minðF Pj Þ  F Pi j ¼ 1; . . . ; ng
where F fi is the fitness value of ith individual. A numerical simula-
tion of an earthquake excited 10-story building is carried out and
the proposed algorithm is able to achieve improved performance
with less control effort.
An active control of structures under wind excitations using a
multi-level optimal design based on GA is proposed (Li et al.,
2000). The proposed Multi-Level Genetic Algorithm (MLGA) con-
siders the number and position of the actuators and control algo-
rithm as multiple optimization problems. This problem has the
properties of non-linearity, non-continuous, and multi-modal
objective function. In Li (2011), a GA is used to tune the mass,
damping, and stiffness of the MRF absorber.
In Xue et al. (2011), a feedback controller is designed, where the
feedback gains are optimized using a GA. The controller also con-
siders the time-delay in applying control forces to the devices.
Two objective functions are: (1) to reduce the displacement and
acceleration response of the ith floor and (2) to reduce the story
drift response as shown in (58) and (59) respectively.
X
r X
r
a1 ^2 þ a2 €xi ½n
xi ½ n ^ 2
^ ¼1
n ^ ¼1
n
( )
X r
^ j X
jd1 ½n r
^ j
jd2 ½n X r
^ j
jdm ½n
max ; ;...;
^ ¼1
n
xm0 n^¼1 xm0 ^ ¼1
n
xm0
where a1 and a2 are the weights of displacement and acceleration
responses respectively, di ½n^  is the story drift from the ith to
(i  1)th floor at the time data point n^ , and xm0 is the maximum dis-
placement responses in all stories. The effectiveness of the proposed
method is demonstrated using numerical simulation of a 3-story
and 8-story structures excited by different seismic forces.
The disadvantage of the GA is that, it requires long computa-
tional time if the number of variables involved in the computation
increases. A modified GA strategy is proposed in Perry et al. (2006)
to improve the computational time efficiency, which uses the
Search Space Reduction Method (SSRM) using a Modified GA based
on Migration and Artificial Selection (MGAMAS) strategy. In order
to improve the computational performance, the algorithm utilizes
some novel ideas including nonlinear cyclic mutation, tagging, and
reduced data input.
5. Applications of structural vibration control systems
In the past few decades, there have been several applications of
vibration control systems in building structures. The implementa-
tion of control devices in buildings and bridges has been investi-
gated by many researchers worldwide. It is reported that over
16,000 building structures around the world have been protected
using anti-seismic systems (Martelli, Forni, & Panza, 2011). Passive
control, such as seismic isolator, is one of the widely implemented
technique. Japan has about 18% of all earthquakes on the planet, of
magnitude 7 or more. Over 5000 buildings in Japan have already
been protected by seismic isolators. A brief review on the applica-
tion of seismic isolations in Japan can be found in Fujita (1998),
Kawamura, Sugisaki, Ogura, Maezawa, and Tanaka (2000). A recent
survey on the performance evaluation of the seismic isolation is
presented in Buckle and Mayes (2012). It is reported that the
ð56Þ USC University Hospital, world’s first base isolated hospital sur-
vived the 1994 Northridge earthquake (Mag-6.7 caused $20 billion
in damage), without any damage. Some other examples are, the
world’s largest base-isolated computer center survived the 1995
Kobe earthquake (Mag-7.1 caused $150 billion in damage) and
the 7-story Christchurch Women’s Hospital in the South Island of
New Zealand survived the February 2011 Christchurch earthquake
(Mayes, Brown, & Pietra, 2012).
ð57Þ Active devices, such as AMD are implemented to mitigate hu-
man-induced vibrations on the in-service footbridge. With a mass
ratio of 0.2%, the vibration is reduced by 60–80% (Casado, Díaz,
Sebastián, V Poncela, & Lorenzana, 2013). The Shanghai World
Financial Center in China, holds two AMDs below its observation
floors to reduce the building vibration during windstorms. During
earthquakes, complete active control is disabled, hence works as
TMDs (Lu, Li, Guo, Shi, & Liu, 2012). The first full-scale implemen-
tation of semi-active MR damper is applied to the Miraikan Build-
ing in Tokyo (Dong & Yuan, 2012). The Osaka Applause Tower in
Osaka, Japan is equipped with an AMD. This damper system has
attenuated the structural response due to the typhoon on July
26th, 1996 and earthquake on April 22nd, 1998 (Dong & Yuan,
2012). Some practical applications of active and semi-active vibra-
tion control of buildings are reported in Casciati, Rodellar, and
Yildirim (2012), Ikeda (2009). The Island Tower Sky Club in Fukuo-
ka City, Japan, uses different kinds of dampers for protection
against wind and seismic forces. A hybrid base isolation system
is applied to reduce the building response during large earth-
quakes. Its towers are interconnected using dampers to reduce
the overturning effect (Nishimura et al., 2011). A state-of-the-art-
ð58Þ
review of the behavior of isolated bridges to seismic excitation is
presented in Kunde and Jangid (2003). More information on build-
ð59Þ ings with vibration control systems can be found in Datta (2003),
Housner et al. (1997), Soong and Spencer (2002), Spencer and Sain
(1997).
On March 11th, 2011, a mega-thrust earthquake of moment
magnitude 9.0 occurred offshore NE Japan. This is the most devas-
tating earthquake in Japan after the 1923 Great Kanto earthquake.
The building experienced strong shaking due to the long-period
ground motions (0.015–0.1 Hz). Despite these conditions, it is re-
ported that many buildings equipped with structural control sys-
tem has performed well. Many research results about this
earthquake are presented in ‘‘International symposium on engi-
neering lessons learned from the 2011 Great East Japan Earth-
quake, UC Berkeley, March 1–4, 2012’’. In Kasai, Pu, and Wada
(2012), the response of eight buildings with structural control sys-
tems are reported and found that the control system responded
well. It states that a 54-story steel building experienced a maxi-
mum displacement of 0.50 m, which would have reached
0.70 min the absence of the control. The Ishinomaki Red Cross Hos-
pital in Ishinomaki, situated near to the epicenter, had a good con-
trolled response (Mayes et al., 2012). Miwada, Yoshida, Ishikawa,
Nakamura, and (2012) shows that the base isolated systems using
high damping rubber bearings in Miyagi and Chiba prefectures
performed well during the Tohoku earthquake. Takewaki, Muraka-
mi, Fujita, Yoshitomi, and Tsuji (2011) shows that some super
high-rise buildings with viscoelastic dampers like high-hardness
rubber dampers controlled the floor vibration effectively.
Beside these successful performances, some buildings with
anti-seismic systems are damaged during Tohoku-Oki earthquake.
A damage to a base isolated building with a lead damper is indi-
cated in Motosaka and Mitsuji (2012). It is also reported that the
structures designed by post-1990 code are few damaged by the
ground motion, whereas many structures designed by post-1995
code, using rubber bearings and dampers are severely damaged
(Takahashi, 2012).
 S. Thenozhi, W. Yu / Annual Reviews in Control 37 (2013) 346–364
Also some control system did not respond to this earthquake.
For example, the damper system (three TMDs) in Taipei 101 sky-
scraper acted during the 2005 Typhoon Long Wang, the 2008
Wenchuan Earthquake, and the 2010 Typhoon Fanapi. However
it did not respond during the 2011 Tohoku-Oki earthquake, be-
cause it was not excited under those circumstances (Chen, Wang,
Huang, Liu, & Huang, 2013). From an engineering point of view,
the observations of these events will be useful for researchers to
identify relevant research questions about the structure safety
and for improving structure resilience against these natural haz-
ards and it is also important to investigate the performance of
the anti-seismic systems to different excitations.
6. Conclusions
In this review, vibration control of structures under both the
earthquake and wind excitations is considered. It address the
developments in modeling and control of building structures, and
tries to include all possible technical aspects of structural control
systems and building modeling. The research in this field is still
growing with new type of control devices and their configurations
and with new control strategies. The emphasis is given to the cur-
rent developments in control strategies in the last two decades,
which shows its significant improvements. This paper concludes
with some important observations as follows:
(1) Most of the structural control reviews emphasize more on
control devices than the control strategies. This paper stud-
ies the vibration control of building structures from the
point of control theory, hence focuses more on controllers
and modeling techniques.
(2) In recent years, semi-active and hybrid control got more
attention. The MR dampers are the commonly used control
device.
(3) Stability is an important criterion in control design. Only a
few structural controllers, such as H1 and SMC, consider
the stability, whereas the other control strategies do not.
(4) Some control algorithms like LQR needs the knowledge of
system parameters, on the other hand techniques like intel-
ligent controls do not.
(5) Time-delay in the controlling is not considered in most of
the work, which is critical in closed-loop system stability.
(6) The actuator saturation has never been discussed, which is
more important during practical implementation.
(7) The optimal placement of the devices is not studied in most
of the papers reviewed.
(8) Acceleration signals are the most reliable source during seis-
mic events, which includes noise and offset. Most of the con-
trollers use displacement and velocity as system states,
which are not easy to obtain from the acceleration signal
with simple integration.
(9) The structural control is normally realized by a central com-
putational device, which may fail during the seismic event.
That will eventually result in total control system collapse.
Decentralized system is a good solution to this problem
(Nezhad & Rofooei, 2007).
(10) Some safety measurements need improvement to incorpo-
rate the effects of long period ground motions in high-rise
buildings.
Acknowledgments
The authors would like to thank Dr. Rubén Garrido for his valu-
able suggestions on this paper. The first author would like to thank
361
Consejo Nacional de Ciencia y Tecnología (CONACyT) of Mexico for
the financial support. This work was partially supported by State
Key Laboratory of Synthetical Automation for Process Industries
and the Project 111 (No. B08015) of China.
References
Adhikari, R., & Yamaguchi, H. (1997). Sliding mode control of buildings with ATMD.
Earthquake Engineering and Structural Dynamics, 26, 409–422.
Adhikari, R., Yamaguchi, H., & Yamazaki, T. (1998). Modal space sliding-mode
control of structures. Earthquake Engineering and Structural Dynamics, 27,
1303–1314.
Agrawal, A. K., Fujino, Y., & Bhartia, B. K. (1993). Instability due to time delay and its
compensation in active control of structures. Earthquake Engineering and
Structural Dynamics, 22, 211–224.
Agrawal, A. K., & Yang, J. N. (1997). Effect of fixed time delay on stability and
performance of actively controlled civil engineering structures. Earthquake
Engineering and Structural Dynamics, 26, 1169–1185.
Agrawal, A. K., & Yang, J. N. (2000). Compensation of time-delay for control of civil
engineering structures. Earthquake Engineering and Structural Dynamics, 29,
37–62.
Ahmadizadeh, M. (2007). On equivalent passive structural control systems for semi-
active control using viscous fluid dampers. Structural Control and Health
Monitoring, 14, 858–875.
Alavinasab, A., & Moharrami, H. (2006). Active control of structures using energy-
based LQR method. Computer-Aided Civil and Infrastructure Engineering, 21,
605–611.
Aldawod, M., Naghdy, F., Samali, B., & Kwok, K. C. S. (1999). Active control of wind
excited structures using fuzzy logic. IEEE International Fuzzy Systems Conference
Proceedings, 72–77.
Ali, S. F., & Ramaswamy, A. (2009). Optimal fuzzy logic control for MDOF structural
systems using evolutionary algorithms. Engineering Applications of Artificial
Intelligence, 22, 407–419.
Allen, M., Zazzera, F. B., & Scattolini, R. (2000). Sliding mode control of a large
flexible space structure. Control Engineering Practice, 8, 861–871.
Alli, H., & Yakut, O. (2005). Fuzzy sliding-mode control of structures. Engineering
Structures, 27, 277–284.
Amini, F., & Tavassoli, M. R. (2005). Optimal structural active control force, number
and placement of controllers. Engineering Structures, 27, 1306–1316.
Åström, K. J., & Eykhoff, P. (1971). System identification – A survey. Automatica, 7,
123–162.
Balendra, T., Wang, C. M., & Yan, N. (2001). Control of wind-excited towers by active
tuned liquid column damper. Engineering Structures, 23, 1054–1067.
Bharti, S. D., Dumne, S. M., & Shrimali, M. K. (2010). Seismic response analysis of
adjacent buildings connected with MR dampers. Engineering Structures, 32,
2122–2133.
Boada, M. J. L., Calvo, J. A., Boada, B. L., & Díaz, V. (2011). Modeling of
amagnetorheologicaldamper by recursivelazylearning. International Journal of
Non-Linear Mechanics, 46, 479–485.
Boore, D. M. (2003). Analog-to-digital conversion as a source of drifts in
displacements derived from digital recordings of ground acceleration. Bulletin
of the Seismological Society of America, 93, 2017–2024.
Brokate, M., & Visintin, A. (1989). Properties of the Preisach model for hysteresis.
Journal für die Reine und Angewandte Mathematik, 402, 1–40.
Buckle, I. G., & Mayes, R. L. (2012). Seismic isolation: History, application, and
performance – A world view. Earthquake spectra, 6, 161–201.
Cao, H. (1997). Analysis and design of active tuned mass damper systems. Ph.D.
Dissertation, State University of New York, Buffalo, USA.
Cao, H., & Li, Q. S. (2004). New control strategies for active tuned mass damper
systems. Computers and Structures, 82, 2341–2350.
Casado, C. M., Díaz, I. M., Sebastián, J. D., V Poncela, A., & Lorenzana, A. (2013).
Implementation of passive and active vibration control on an in-service
footbridge’’. Structural Control And Health Monitoring, 20, 70–87.
Casciati, F., Rodellar, J., & Yildirim, U. (2012). Active and semi-active control of
structures – Theory and applications: A review of recent advances. Journal of
Intelligent Material Systems and Structures, 23(11), 1181–1195.
Cervantes, J. M. A., & Icaza, L. A. (2010). Identification of seismically excited
buildings with two orthogonal horizontal components. Journal of Vibration and
Control, 17, 881–901.
Chang, C. C. (1999). Mass dampers and their optimal designs for building vibration
control. Engineering Structures, 21, 454–463.
Chang, C. M., & Spencer, B. F. (2010). Active base isolation of buildings subjected to
seismic excitations. Earthquake Engineering and Structural Dynamics, 39,
1493–1512.
Chang, J. C. H., & Soong, T. T. (1980). Structural control using active tuned mass
damper. Journal of Engineering Mechanics, ASCE, 106, 1091–1098.
Chassiakos, A. G., & Masri, S. F. (1996). Identification of structural systems by neural
networks. Mathematics and Computers in Simulation, 40, 637–656.
Chen, C. W., Chiang, W. L., Hsiao, F. H., & Tsai, C. H. (2004). H1 fuzzy control of
structural systems using Takagi–Sugeno fuzzy model. Proceedings of the IEEE
International Conference on Mechatronics, 340–345.
Chen, C. W. (2009). Modeling and control for nonlinear structural systems via a NN-
based approach. Expert Systems with Applications, 36, 4765–4772.
362 S. Thenozhi, W. Yu / Annual Reviews in Control 37 (2013) 346–364
Chen, K. C., Wang, J. H., Huang, B. S., Liu, C. C., & Huang, W. G. (2013). Vibrations of
the TAIPEI 101 skyscraper caused by the 2011 Tohoku earthquake, Japan. Earth
Planets Space, 64, 1277–1286.
Cheng, F. Y., Jiang, H., & Lou, K. (2008). Smart structures: Innovative systems for
seismic response control. CRC Press.
Cho, H. C., Fadali, M. S., Saiidi, M. S., & Lee, K. S. (2005). Neural network active
control of structures with earthquake excitation. International Journal of Control,
Automation and Systems, 2, 202–210.
Choi, K. M., Cho, S. W., Kim, D. O., & Lee, I. W. (2005). Active control for seismic
response reduction using modal-fuzzy approach. International Journal of Solids
and Structures, 42, 4779–4794.
Chopra, A. K. (2001). Dynamics of structures: Theory and application to earthquake
engineering (2nd ed.). Prentice Hall.
Chua, L. O., & Bass, S. C. (1972). A generalized hysteresis model. IEEE Transaction of
Circuit Theory, 19, 36–48.
Chung, L. L., Wu, L. Y., & Jin, T. G. (1998). Acceleration feedback control of seismic
structures. Engineering Structures, 20, 62–74.
Lord Corporation. (1995). Versaflo product information sheet. Lord Corporation
Publication, No. PI02-MRX-135CD.
Das, D., Datta, T. K., & Madan, A. (2012). Semiactive fuzzy control of the seismic
response of building frames with MR dampers. Earthquake Engineering and
Structural Dynamics, 41, 99–118.
Datta, T. K. (2003). A state-of-the-art review on active control of structures. ISET
Journal of Earthquake Technology, 40, 1–17.
Djajakesukma, S. L., Samali, B., & Nguyen, H. (2002). Study of a semi-active stiffness
damper under various earthquake inputs. Earthquake Engineering and Structural
Dynamics, 31, 1757–1776.
Dong, Z. H., & Yuan, J. Y. (2012). Vibration control device and its performance under
wind load in high-rise buildings. Applied Mechanics and Materials, 166–169,
1358–1361.
Du, H., & Zhang, N. (2008). H1 control for buildings with time delay in control via
linear matrix inequalities and genetic algorithms. Engineering Structures, 30,
81–92.
Duc, N. D., Vu, N. L., Tran, D. T., & Bui, H. L. (2011). A study on the application of
hedge algebras to active fuzzy control of a seism-excited structure. Journal of
Vibration and Control, 1–15.
Dyke, S. J., Spencer, B. F., Quast, P., Sain, M. K., Kaspari, D. C., & Soong, T. T. (1996).
Acceleration feedback control of MDOF structures. Journal of Engineering
Mechanics, 122, 907–918.
Dyke, S. J., Spencer, B. F., Quast, P., Kaspari, D. C., & Sain, M. K. (1996).
Implementation of an active mass driver using acceleration feedback control.
Computer-Aided Civil and Infrastructure Engineering, 11, 305–323.
Edwards, T. S. (2007). Effects of aliasing on numerical integration. Mechanical
Systems and Signal Processing, 21, 165–176.
Fabián, R. J., & Icaza, L. A. (2010). An adaptive observer for a shear building with an
energy-dissipation device. Control Engineering Practice, 18, 331–338.
Fisco, N. R., & Adeli, H. (2011). Smart structures: Part I – Active and semi-active
control. Scientia Iranica, 18, 275–284.
Fisco, N. R., & Adeli, H. (2011). Smart structures: Part II – Hybrid control systems
and control strategies. Scientia Iranica, 18, 285–295.
Flynn, E. B., & Todd, M. D. (2010). A Bayesian approach to optimal sensor placement
for structural health monitoring with application to active sensing. Mechanical
Systems and Signal Processing, 24, 891–903.
Fleming, P. J., & Purshouse, R. C. (2002). Evolutionary algorithms in control systems
engineering: A survey. Control Engineering Practice, 10, 1223–1241.
Forrai, A., Hashimoto, S., Funato, H., & Kamiyama, K. (2001). Structural control
technology: system identification and control of flexible structures. Computing
and control engineering Journal, 402, 1–40.
Fujita, T. (1998). Seismic isolation of civil buildings in Japan. Progress in Structural
Engineering and Materials, 1(3), 295–300.
Garrido, R., & Francisco, J. R. (2006). Hysteresis and parameter estimation of MDOF
systems by a continuous-time least squares method. Journal of Earthquake
Engineering, 10, 237–264.
Gavin, H. P., Morales, R., & Reilly, K. (1998). Drift-free integrators. Review of Scientific
Instruments, 69, 2171–2175.
Gawronski, W. (1997). Actuator and sensor placement for structural testing and
control. Journal of Sound and Vibration, 208, 101–109.
Gu, Z. Q., & Oyadiji, S. O. (2008). Application of MR damper in structural control
using ANFIS method. Computers and Structures, 86, 427–436.
Guclu, R. (2006). Sliding mode and PID control of a structural system against
earthquake. Mathematical and Computer Modelling, 44, 210–217.
Guclu, R., & Yazici, H. (2008). Vibration control of a structure with ATMD against
earthquake using fuzzy logic controllers. Journal of Sound and Vibration, 318,
36–49.
Guney, M., & Eskinat, E. (2008). Optimal actuator and sensor placement in flexible
structures using closed-loop criteria. Journal of Sound and Vibration, 312,
210–233.
Hiramoto, K., Doki, H., & Obinata, G. (2000). Optimal sensor/actuator placements for
active vibration control using explicit solution of algebraic Riccati equation.
Journal of Sound and Vibration, 229, 1057–1075.
Ho, C. C., & Ma, C. K. (2007). Active vibration control of structural systems by a
combination of the linear quadratic Gaussian and input estimation approaches.
Journal of Sound and Vibration, 301, 429–449.
Hong, A. L., Betti, R., & Lin, C. C. (2009). Identification of dynamic models of a
building structure using multiple earthquake records. Structural Control and
Health Monitoring, 16, 178–199.
Housner, G. W., Bergman, L. A., Caughey, T. K., Chassiakos, A. G., Claus, R. O., Masri, S.
F., et al. (1997). Present and future. Journal of Engineering Mechanics, 123,
897–974.
Housner, G. W., Soong, T. T., & Masri, S. F. (1996). Second generation of active
structural control in civil engineering. Computer-Aided Civil and Infrastructure
Engineering, 11, 289–296.
Hrovat, D., Barak, P., & Rabins, M. (1983). Semi-active versus passive or active tuned
mass dampers for structural control. Journal of Engineering Mechanics, 109,
691–705.
Hung, S. L., Huang, C. S., Wen, C. M., & Hsu, Y. C. (2003). Nonparametric
identification of a building structure from experimental data using wavelet
neural network. Computer-Aided Civil and Infrastructure Engineering, 18,
356–368.
Ikeda, Y. (2009). Active and semi-active vibration control of buildings in Japan –
Practical applications and verification. Structural Control And Health Monitoring,
16, 703–723.
Ikhouane, F., & Rodellar, J. (2007). Systems with hysteresis: Analysis, identification and
control using the Bouc–Wen model. Chichester: Wiley.
Imai, H., Yun, C. B., Maruyama, O., & Shinozuka, M. (1989). Fundamentals of system
identification in structural dynamics. Probabilistic Engineering Mechanics, 4,
162–173.
Iuliis, M. D., & Faella, C. (2013). Effectiveness analysis of a semiactive base isolation
strategy using information from an early-warning network. Engineering
Structures, 52, 518–535.
Jiang, B., Wei, X., & Guo, Y. (2010). Linear quadratic optimal control in active control
of structural vibration systems. In Control and decision conference, 2010 Chinese
(vol. 98, pp. 3546–3551).
Jones, N. P., Shi, T., Ellis, J. H., & Scanlan, R. H. (1995). System-
identification procedure for system and input parameters in ambient
vibration surveys. Journal of Wind Engineering and Industrial Aerodynamics, 54-
55, 91–99.
Kasai, K., Pu, W., & Wada, A. (2012). Responses of controlled tall buildings in Tokyo
subjected to the Great East Japan earthquake. In Proceedings of the international
symposium on engineering lessons learned from the 2011 Great East Japan
earthquake (pp. 1099–1109).
Kawamura, S., Sugisaki, R., Ogura, K., Maezawa, S., & Tanaka, S. (2000). Seismic
isolation retrofit in Japan. In Proceedings of the 2th world conference on
earthquake engineering (pp. 1–8).
Kelly, T. E. (2001). Base isolation of structures: Design guidelines. Holmes Consulting
Group Ltd.
Kerschen, G., Worden, K., Vakakis, A. F., & Golinva, J. C. (2006). Past, present and
future of nonlinear system identification in structural dynamics. Mechanical
Systems and Signal Processing, 20, 505–592.
Kharaajoo, M. J., Kangharloo, K., & Roudsari, F. H. (2004). Fuzzy sliding mode control
design based on genetic algorithms: Application to building structures. IEEE
International Conference on Industrial Technology, 27–31.
Kim, S. B., & Yun, C. B. (2000). Sliding mode fuzzy control: Theory and verification on
a benchmark structure. Earthquake Engineering and Structural Dynamics, 29,
1587–1608.
Kim, J. T., Jung, H. J., & Lee, I. W. (2000). Optimal structural control using neural
networks. Journal of Engineering Mechanics, 126, 201–205.
Kim, H., & Adeli, H. (2005). Hybrid control of smart structures using a novel
wavelet-based algorithm. Computer-Aided Civil and Infrastructure Engineering,
20, 7–22.
Kim, D. H., Kimb, D., Chang, S., & Jung, H. Y. (2008). Active control strategy of
structures based on lattice type probabilistic neural network. Probabilistic
Engineering Mechanics, 23, 45–50.
Kim, D. H. (2009). Neuro-control of fixed offshore structures under earthquake.
Engineering Structures, 31, 517–522.
Kirsch, U. (2008). Reanalysis of structures: A unified approach for linear, nonlinear,
static, and dynamic systems. Springer.
Korkmaz, S. (2011). A review of active structural control: Challenges for engineering
informatics. Computers and Structures, 89, 2113–2132.
Krasnoselskii, M. A., & Pokrovskii, A. V. (1989). Systems with hysteresis. New York:
Springer-Verlag.
Kunde, M. C., & Jangid, R. S. (2003). Seismic behavior of isolated bridges: A-state-of-
the-art review. Electronic Journal of Structural Engineering, 3, 140–170.
Kwok, K. C. S., & Samali, B. (1995). Performance of tuned mass dampers under wind
loads. Engineering Structures, 17, 655–667.
Lee, C. G., & Yun, C. B. (1991). Parameter identification of linear structural dynamic
systems. Computers and Structures, 40, 1475–1487.
Li, L. J. (2011). MRF absorber damping control for building structural vibration
response by means of genetic optimum algorithm. Advanced Materials Research,
219-220, 1133–1137.
Li, Q. S., Liu, D. K., Fang, J. Q., & Tam, C. M. (2000). Multi-level optimal design of
buildings with active control under winds using genetic algorithms. Journal of
Wind Engineering and Industrial Aerodynamics, 86, 65–86.
Li, Z. N., Tang, J., & Li, Q. S. (2004). Optimal sensor locations for structural vibration
measurements. Applied Acoustics, 65, 807–818.
Li, Z., Deng, Z., & Gu, Z. (2010). New sliding mode control of building structure using
RBF neural networks. Chinese Control and Decision Conference, 2820–2825.
Liang, Z., Lee, G. C., Dargush, G. F., & Song, J. (2011). Structural damping: Applications
in seismic response modification. CRC Press.
Lin, L., Dyke, S. J., & Rebecca, V. (2007). Wireless sensing and control of structural
vibration from earthquake. In Proceedings of the 26th Chinese control conference,
Hunan, China (pp. 194–198).
 S. Thenozhi, W. Yu / Annual Reviews in Control 37 (2013) 346–364
Lin, C. C., Hong, L. L., Ueng, J. M., Wu, K. C., & Wang, C. E. (2005). Parametric
identification of asymmetric buildings from earthquake response records. Smart
Materials and Structures, 14, 850–861.
Link, A., & von Martens, H. J. (2004). Accelerometer identification using shock
excitation. Measurement, 35, 191–199.
Liu, D. K., Yang, Y. L., & Li, Q. S. (2003). Optimum positioning of actuators in tall
buildings using genetic algorithm. Computers and Structures, 81, 2823–2827.
Liu, J., Xia, K., & Zhu, C. (2009). Structural vibration intelligent control based on
magnetorheological damper. International Conference on Computational
Intelligence and Software Engineering, 1–4.
Liu, W., Hou, Z., & Demetriou, M. A. (2006). A computational scheme for the optimal
sensor/actuator placement of flexible structures using spatial H2 measures.
Mechanical Systems and Signal Processing, 20, 881–895.
Ljung, L. (1987). System identification theory for the users. New Jersey: Prentice-Hall,
Inc.
Lu, X., Li, P., Guo, X., Shi, W., & Liu, J. (2012). Vibration control using ATMD and site
measurements on the Shanghai World Financial Center Tower. The Structural
Design of Tall and Special Buildings.
Mackriell, L. E., Kwok, K. C. S., & Samali, B. (1997). Critical mode control of a wind-
loaded tall building using an active tuned mass damper. Engineering Structures,
19, 834–842.
Madan, A. (2005). Vibration control of building structures using self-organizing and
self-learning neural networks. Journal of Sound and Vibration, 287, 759–784.
Mandic, D. P., & Chambers, J. A. (2001). Recurrent neural networks for prediction:
Learning algorithms, architectures and stability. John Wiley and Sons.
Martelli, A., Forni, M., & Panza, G. (2011). Features, recent application and
conditions for the correct use of seismic isolation systems. Seismic Control
Systems: Design and Performance Assessment, 120, 15–27.
Marzbanrad, J., Ahmadi, G., & Jha, R. (2004). Optimal preview active control of
structures during earthquakes. Engineering Structures, 26, 1463–1471.
Mayergoyz, I. D. (1991). Mathematical models of hysteresis. New York: Springer-
Verlag.
Mayes, R. L., Brown, A. G., & Pietra, D. (2012). Using seismic isolation and energy
dissipation to create earthquake-resilient buildings. Bulletin of the New Zealand
Society for Earthquake Engineering, 45, 117–122.
McNamara, R. J. (1977). Tuned mass dampers for buildings. Journal of the Structural
Division, 103, 1785–1798.
Min, K. W., Kim, H. S., Lee, S. H., Kim, H., & Ahn, S. K. (2005). Performance evaluation
of tuned liquid column dampers for response control of a 76-story benchmark
building. Engineering Structures, 27, 1101–1112.
Miwada, G., Yoshida, O., Ishikawa, R., & Nakamura,, M. (2012). Observation records
of base-isolated buildings in strong motion area during the 2011 off the pacific
coast of tohoku earthquake. In Proceedings of the international symposium on
engineering lessons learned from the 2011 Great East Japan earthquake (pp. 1017–
1024).
Motosaka, M., & Mitsuji, K. (2012). Building damage during the 2011 off the Pacific
coast of Tohoku earthquake. Soils and Foundations, 52(5), 929–944.
Nerves, A. C., & Krishnan, R. (1995). Active control strategies for tall civil structures.
Proceedings of IEEE, International Conference on Industrial Electronics, Control, and
Instrumentation, 2, 962–967.
Nezhad, S. M., & Rofooei, F. R. (2007). Decentralized sliding mode control of
multistory buildings. The Structural Design of Tall and Special Buildings, 16,
181–204.
Nishimura, A., Yamamoto, H., Kimura, Y., Kimura, H., Yamamoto, M., & Kushibe, A.
(2011). Base-isolated super high-rise RC building composed of three connected
towers with vibration. Structural Concrete, 12(2), 94–108.
Obe, O. I. (1985). Optimal actuators placements for the active control of flexible
structures. Journal of Mathematical Analysis and Applications, 105, 12–25.
Pandya, J., Akbay, Z., Uras, M., & Aktan, H. (1996). Experimental implementation of
hybrid control. In Proceedings of structures congress XIV, Chicago (pp. 1172–
1179).
Park, K. S., & Koh, H. M. (2004). Preference-based optimum design of an integrated
structural control system using genetic algorithms. Advances in Engineering
Software, 35, 85–94.
Park, K. S., Koh, H. M., & Ok, S. Y. (2002). Active control of earthquake excited
structures using fuzzy supervisory technique. Advances in Engineering Software,
33, 761–768.
Park, K. S., Koh, H. M., & Seo, C. W. (2004). Independent modal space fuzzy control of
earthquake-excited structures. Engineering Structures, 26, 279–289.
Park, K. T., Kim, S. H., Park, H. S., & Lee, K. W. (2005). The determination of bridge
displacement using measured acceleration. Engineering Structures, 27,
371–378.
Park, K. S., Kohb, H. M., Okb, S. Y., & Seo, C. W. (2005). Fuzzy supervisory control of
earthquake-excited cable-stayed bridges. Engineering Structures, 27, 1086–1100.
Park, W., Ha, D. H., Park, K. S., & Choo, J. F. (2009). Preference based genetic
algorithm for the optimum design of integrated structural control system. In
IEEE, Fifth international joint conference on INC, IMS and IDC (pp. 389–392).
Park, W., Park, K. S., & Koh, H. M. (2008). Active control of large structures using a
bilinear pole-shifting transform with H1 control method. Engineering Structures,
30, 3336–3344.
Peeters, B., & Roeck, G. D. (2001). Stochastic system identification for operational
modal analysis: A review. Journal of Dynamic Systems, Measurement, and Control,
123, 659–667.
Perry, M. J., Koh, C. G., & Choo, Y. S. (2006). Modified genetic algorithm strategy for
structural identification. Computers and Structures, 84, 529–540.
363
Pourzeynali, S., Lavasani, H. H., & Modarayi, A. H. (2007). Active control of high rise
building structures using fuzzy logic and genetic algorithms. Engineering
Structures, 29, 346–357.
Quek, S. T., Wang, W., & Koh, C. G. (1999). System identification of linear MDOF
structures under ambient excitation. Earthquake Engineering and Structural
Dynamics, 28, 61–77.
Quiñonero, F. P., Massegú, J. R., Rossell, J. M., & Karimi, H. R. (2012). Semiactive-
passive structural vibration control strategy for adjacent structures under
seismic excitation’’. Journal of the Franklin Institute, 349, 3003–3026.
Razavi, S. H., Abolmaali, A., & Ghassemieh, M. (2007). A weighted residual parabolic
acceleration time integration method for problems in structural dynamics.
Journal of Computational Methods in Applied Mathematics, 7, 227–238.
Reithmeier, E., & Leitmann, G. (2001). Structural vibration control. Journal of the
Franklin Institute, 338, 203–223.
Ribeiro, J. G. T., de Castro, J. T. P., & Freire, J. L. F. (2003). Using the FFT-DDI method to
measure displacements with piezoelectric, resistive and ICP accelerometers. In
Conference and exposition on structural dynamics.
Sain, P. M., Sain, M. K., & Spencer, B. F. (1997). Models for hysteresis and application
to structural control. Proceedings of the American Control Conference, 16–20.
Saragih, R. (2010). Designing active vibration control with minimum order for
flexible structures. IEEE International Conference on Control and Automation,
450–453.
Seto, K. (1996). A structural control method of the vibration of flexible buildings in
response to large earthquake and strong winds. In Proceedings of the 35th
conference on decision and control, Kobe, Japan.
Shames, I. H., & Cozzarelli, F. A. (1992). Elastic and inelastic stress analysis. Englewood
Cliffs: Prentice Hall.
Shook, D. A., Roschke, P. N., Lin, P. Y., & Loh, C. H. (2008). GA-optimized fuzzy logic
control of a large-scale building for seismic loads. Engineering Structures, 30,
436–449.
Singh, M. P., & Matheu, E. E. (1997). Active and semi-active control of structures
under seismic excitation. Earthquake Engineering and Structural Dynamics, 26,
193–213.
Soong, T. T. (1990). Active structural control: Theory and practice. New York:
Longman.
Soong, T. T., Reinhorn, A. M., Wang, Y. P., & Lin, R. C. (1991). Full-scale
implementation of active control-I: Design and simulation. Journal of
Structural Engineering, 117, 3516–3536.
Soong, T. T., Masri, S. F., & Housner, G. W. (1991). An overview of active structural
control under seismic loads. Earthquake Spectra, 7, 483–505.
Soong, T. T., & Spencer, B. F. (2002). Supplemental energy dissipation: State-of-the-
art and state-of-the-practice. Engineering Structures, 24, 243–259.
Spencer, B. F. (1986). Reliability of randomly excited hysteretic structures. New York:
Springer-Verlag.
Spencer, B. F., & Sain, M. K. (1997). Controlling buildings: A new frontier in feedback.
IEEE Control Systems Magazine on Emerging Technology, 17, 19–35.
Spencer, B. F., Dyke, S. J., Sain, M. K., & Carlson, J. D. (1997). Phenomenological model
of a magnetorheological damper. Journal of Engineering Mechanics, ASCE, 123,
230–238.
Spencer, B. F., & Nagarajaiah, S. (2003). State of the art of structural control. Journal
of Structural Engineering, 129, 845–856.
Symans, M. D., & Constantinou, M. C. (1996). Semi-active control of earthquake
induced vibration. In Eleventh world conference on earthquake engineering, Paper
No: 95.
Symans, M. D., & Constantinou, M. C. (1999). Semi-active control systems for
seismic protection of structures: A state-of-the-art review. Engineering
Structures, 21, 469–487.
Takahashi, Y. (2012). Damage of rubber bearings and dampers of bridges in 2011
Great East Japan earthquake. In Proceedings of the international symposium on
engineering lessons learned from the 2011 Great East Japan earthquake (pp. 1333–
1342).
Takewaki, I., Murakami, S., Fujita, K., Yoshitomi, S., & Tsuji, M. (2011). The 2011 off
the Pacific coast of Tohoku earthquake and response of high-rise buildings
under long-period ground motions. Soil Dynamics and Earthquake Engineering,
31(11), 1511–1528.
Tang, Y. (1996). Active control of SDF systems using artificial neural networks.
Computers and Structures, 60, 695–703.
Tani, A., Kawamura, H., & Ryu, S. (1998). Intelligent fuzzy optimal control of
building structures. Engineering Structures, 20, 184–192.
Teng, T. L., Peng, C. P., & Chuang, C. (2000). A study on the application of fuzzy
theory to structural active control. Computer Methods in Applied Mechanics and
Engineering, 189, 439–448.
Thenozhi, S., Yu, W., & Garrido, R. (2013). A novel numerical integrator for velocity
and position estimation. Transactions of the Institute of Measurement and Control,
35, 824–833.
Thong, Y. K., Woolfson, M. S., Crowe, J. A., Gill, B. R. H., & Jones, D. A. (2004).
Numerical double integration of acceleration measurements in noise.
Measurement, 36, 73–92.
Tinkir, M., Kalyoncu, M., & Sßahin, Y. (2013). Deflection control of two-floors
structure against northridge earthquake by using PI controlled active mass
damping. Applied Mechanics and Materials, 307, 126–130.
Utkin, V. I. (1990). Sliding modes in control and optimization. Berlin: Springer-Verlag.
Wang, A. P., & Lee, C. D. (2002). Fuzzy sliding mode control for a building structure
based on genetic algorithms’’. Earthquake Engineering and Structural Dynamics,
31, 881–895.
364 S. Thenozhi, W. Yu / Annual Reviews in Control 37 (2013) 346–364
Wang, A. P., & Lin, Y. H. (2007). Vibration control of a tall building subjected to
earthquake excitation. Journal of Sound and Vibration, 299, 757–773.
Wang, J. Y., Ni, Y. Q., Ko, J. M., & Spencer, B. F. (2005). Magneto-rheological tuned
liquid column dampers (MR-TLCDs) for vibration mitigation of tall buildings:
Modelling and analysis of open-loop control. Computers and Structures, 83,
2023–2034.
Wang, S. G., Roschke, P. N., & Yeh, H. Y. (2002). Simulation of robust control for
uncertain structural systems against earthquakes. In Proceedings of the 4th world
congress on intelligent control and automation, Shanghai, PR China (pp. 1681–
1686).
Weiss, K. D., Duclos, T. G., Carlson, J. D., Chrzan, M. J., & Margida, A. J. (1993). High
strength magneto- and electrorheological fluids, society of automotive engineers,
SAE Paper No. 932451.
Wen, Y. K. (1976). Method for random vibration of hysteretic systems. Journal of
Engineering Mechanics, 102, 249–263.
Wen, Y. K., Ghaboussi, J., Venini, P., & Nikzad, K. (1995). Control of structures using
neural networks. Smart Materials and Structures, 4, 149–157.
Worden, K. (1990). Data processing and experiment design for the restoring force
surface method, Part I: Integration and differentiation of measured time data.
Mechanical Systems and Signal Processing, 4, 295–319.
Xu, B., Wu, Z., Chen, G., & Yokoyama, K. (2004). Direct identification of structural
parameters from dynamic responses with neural networks. Engineering
Applications of Artificial Intelligence, 17, 931–943.
Xu, Y. L. (1996). Parametric study of active mass dampers for wind-excited tall
buildings. Engineering Structures, 18, 64–76.
Xu, Z., Agrawal, A. K., & Yang, J. N. (2006). Semi-active and passive control of the
phase I linear base-isolated benchmark building model. Structural Control and
Health Monitoring, 13, 626–648.
Xu, Z. D., & Guo, Y. Q. (2008). Neuro-fuzzy control strategy for earthquake-excited
nonlinear magnetorheological structures. Soil Dynamics and Earthquake
Engineering, 28, 717–727.
Xue, X. M., Sun, Q., Wu, X. H., & Zhang, L. (2011). Study on structural vibration
control by GA controller involving time delay. Applied Mechanics and Materials,
143–144, 335–340.
Yakut, O., & Alli, H. (2011). Neural based sliding-mode control with moving sliding
surface for the seismic isolation of structures. Journal of Vibration and Control,
17, 2103–2116.
Yalla, S. K., Kareem, A., & Kantor, J. C. (2001). Semi-active tuned liquid column
dampers for vibration control of structures. Engineering Structures, 23,
1469–1479.
Yagiz, N. (2001). Sliding mode control of a multi-degree-of-freedom structural
system with active tuned mass damper. Turk Journal of Engineering and
Environmental Science, 25, 651–657.
Yan, G., & Zhou, L. L. (2006). Integrated fuzzy logic and genetic algorithms for multi-
objective control of structures using MR dampers. Journal of Sound and
Vibration, 296, 368–382.
Yang, G., Spencer, B. F., Jr., Carlson, J. D., & Sain, M. K. (2002). Large-scale MR fluid
dampers: modeling and dynamic performance considerations. Engineering
Structures, 24, 309–323.
Yang, J., Li, J. B., & Lin, G. (2006). A simple approach to integration of acceleration
data for dynamic soil–structure interaction analysis. Soil Dynamics and
Earthquake Engineering, 26, 725–734.
Yang, J. N., & Soong, T. T. (1988). Recent advances in active control of civil
engineering structures. Probabilistic Engineering Mechanics, 3, 179–188.
Yang, J. N., Wu, J. C., Agrawal, A. K., & Hsu, S. Y. (1997). Sliding mode control with
compensator for wind and seismic response control. Earthquake Engineering and
Structural Dynamics, 26, 1137–1156.
Yao, J. T. P. (1972). Concept of structural control. Journal of the Structural Division, 98,
1567–1574.
Yeh, K., Chiang, W. L., & Juang, D. S. (1994). Application of fuzzy control theory in
active control of structures. IEEE Proceeding NAFIPS/IFIS/NASA, 243–247.
Yi, F., & Dyke, S. J. (2000). Structural control systems: Performance assessment.
Proceedings of the American Control Conference, 1, 14–18.
Zhang, J., & Roschke, P. N. (1999). Active control of a tall structure excited by wind.
Journal of Wind Engineering and Industrial Aerodynamics, 83, 209–223.
Zhang, X. (2010). Study on fuzzy control algorithm of high-rise buildings structural
vibration. International Conference on Mechanical and Electrical Technology,
701–704.
Zhang, Z., Koh, C. G., & Duan, W. H. (2010). Uniformly sampled genetic algorithm
with gradient search for structural identification – Part I: Global search.
Computers and Structures, 88, 949–962.
Zhang, Z., Koh, C. G., & Duan, W. H. (2010). Uniformly sampled genetic algorithm
with gradient search for structural identification – Part II: Local search.
Computers and Structures, 88, 1149–1161.
Zhao, B., Lu, X., Wu, M., & Mei, Z. (2000). Sliding mode control of buildings with
base-isolation hybrid protective system. Earthquake Engineering and Structural
Dynamics, 29, 315–326.
Zhu, W. H. (2007). Velocity estimation by using position and acceleration sensors.
IEEE Transactions on Industrial Electronics, 54, 2706–2715.
Suresh Thenozhi was born in Palakkad, Kerala, India, in 1985. He received the B.Sc.
and M.Sc. degrees, both in Electronic Science from Bharathiar University, Coimba-
tore, Tamilnadu, India, in 2005 and 2007, respectively and the M.Tech. degree in
Sensor System Technology from VIT University, Vellore, Tamilnadu, India, in 2009.
He is currently doing his Ph.D. degree in the Departamento de Control Automático
at Centro de Investigación y de Estudios Avanzados, National Polytechnic Institute
(CINVESTAV-IPN), Mexico City, Mexico. His current researches are focused on
intelligent control, structural vibration control, robotics, and instrumentation.
Wen Yu received the B.S. degree from Tsinghua University, Beijing, China in 1990
and the M.S. and Ph.D. degrees, both in Electrical Engineering, from Northeastern
University, Shenyang, China, in 1992 and 1995, respectively. From 1995 to 1996, he
served as a Lecturer in the Department of Automatic Control at Northeastern Uni-
versity, Shenyang, China. Since 1996, he has been with the Centro de Investigación y
de Estudios Avanzados, Instituto Politécnico Nacional (CINVESTAV-IPN), Mexico
City, Mexico, where he is currently a Professor with the Departamento de Control
Automatico. From 2002 to 2003, he held research positions with the Instituto
Mexicano del Petroleo. He was a Senior Visiting Research Fellow with Queen’s
University Belfast, Belfast, U.K., from 2006 to 2007, and a Visiting Associate Pro-
fessor with the University of California, Santa Cruz, from 2009 to 2010. He also
holds a visiting professorship at Northeastern University in China from 2006.
Dr.Wen Yu serves as an associate editor of Neurocomputing, and Journal of Intel-
ligent and Fuzzy Systems. He is a member of the Mexican Academy of Sciences.