Download as pdf or txt
Download as pdf or txt
You are on page 1of 14

Received: 5 November 2016 Accepted: 18 November 2016

DOI: 10.1111/exsy.12205

INVITED REVIEW

Recent advances in control algorithms for smart structures and


machines
Mariantonieta Gutierrez Soto | Hojjat Adeli

Department of Civil, Environmental, and


Geodetic Engineering, The Ohio State Abstract
University, Columbus, Ohio 43210, USA This paper presents a state‐of‐the‐art review of recent control algorithms used primarily for
Correspondence vibration control of smart structures and machines. Control algorithms can be divided into open
Hojjat Adeli, Department of Civil, loop, feedback, feedforward, and combined feedback and feedforward control. Solutions pro-
Environmental, and Geodetic Engineering, The
posed can be subdivided into (a) classical control, (b) optimal control, (c) robust control, (d) intel-
Ohio State University, 470 Hitchcock Hall,
2070 Neil Avenue, Columbus, Ohio 43210, ligent control, and (e) self‐sustaining solutions and energy harvesting used in centralized and
USA. decentralized configurations. The most recent research on novel autonomous control systems
Email: adeli.1@osu.edu
intersects distributed control theory, agent‐based modeling, bioinspired computational learning,
and optimization algorithms.

KEY W ORDS

Intelligent control, robust control, smart structures, vibration control

1 | I N T RO D U CT I O N Feedback control systems are more widely used compared with


feedforward control (Vrabie, Vamvoudakis, & Lewis, 2013). A
In general, control laws can be divided into open loop, feedback, feedforward control law is usually implemented when the output
feedforward control, and combination of feedforward and feedback feedback is difficult to obtain and feedback stability is difficult to
control (Figure 1). An open loop system is very simple; an input signal guarantee (Hassani, 2014). To improve the tracking error in control
directs the control element to react, and then an output is produced. problems related to modeling imperfection, dynamics, and uncer-
An example of an open loop system is a washing machine. A person tainties, a combination of feedforward and feedback control structure
selects an input time that sets the timer to get the washing process may be used (Figure 1d).
started, and the output is the washed clothes. This paper presents a state‐of‐the‐art review of recent control
Sometimes, the input signal has to be adjusted based on detect- algorithms with a focus on vibration reduction subdivided into (a) clas-
ing changes in the output using sensor measurements and compari- sical control, (b) optimal control, (c) robust control, (d) intelligent con-
son with a desired result. This dynamic is called feedback control, trol, and (e) self‐sustaining solutions and energy harvesting used in
also known as a closed loop system (Ogata, 2010). An example of centralized and decentralized configurations.
feedback control is an air conditioner system that adjusts the output
of air by comparing the outside temperature with the desired
temperature. 2 | CLASS I CAL CO N TROL
Feedforward control uses the external signal as input to emit an
action of the controller without considering the output of the system, Proportional‐integral‐derivative (PID) control is a feedback control
in other words, without the adjustment characteristic used by a mechanism that continuously calculates an error value between a
feedback control (Adeli & Kim, 2009). In feedforward control algo- desired set point and the measured variable. P stands for the present
rithms, measurements of external disturbances are accounted for values of the error, I accounts for past values of the error, and D
before they have time to affect the system. A feedforward control accounts for possible future values of the error (Dorf & Bishop,
predicts the output signal. For example, feedforward control in a 2008; Ogata, 2010). Even though the use of PID control does not
house can allow the air conditioner to be turned on when a door guarantee optimal control or stability, it is one of the most commonly
opens. In this case, the assumption is that all windows are closed so used control methods in industrial applications by adjusting the tuning
that no air is getting out. parameters (Dorf & Kusiak, 1994).

Expert Systems. 2017;34:e12205. wileyonlinelibrary.com/journal/exsy Copyright © 2017 John Wiley & Sons, Ltd. 1 of 14
https://doi.org/10.1111/exsy.12205
2 of 14 INVITED REVIEW

FIGURE 1 (a) Open loop, (b) feedback, (c)


feedfoward, and (d) combination of feedback
and feedforward control algorithms

2.1 | Feedback control multiple control devices, and considerations should be given to the num-
ber and placement of controlling devices according to the natural fre-
Classical feedback control algorithms have been used to control both
quencies of the structure and also to reduce energy use, size, and cost.
small‐ and large‐scale systems. Moore and Moheimani (2015) use a
Li, Li, Song, and Yu (2014) use positive position feedback control experi-
feedback proportional (P) controller for motion actuation and sensing
mentally for vibration reduction of a one‐dimensional beam with wireless
of a two‐dimensional (2‐D) microelectromechanical (MEM) model in a
and wired sensors. They use time synchronization strategy to create
nanometer scale. The controller is an electrostatic drive that uses volt-
redundancy in wireless network so that when a sensor breaks down,
age to actuate MEM vibrations at resonant frequencies. The drive has
the controller switches to a simulator and predicts the sensor output.
a self‐sensing arrangement that uses the drive's current and charge
measurements as well as MEM's displacement and velocity as output
measurements in the feedback loop to compute the control gain. Sim-
ilarly, Sun, Hu, Moura, and Sun (2015) use a P‐controller to reduce
2.2 | Combined feedback and feedforward control
vibrations of a three degrees of freedom structure platform simulation Rodriguez Fortun, Orus, Alfonso, Gimeno, and Castellanos (2013) com-
subjected to random and periodic external excitations using three‐ bined velocity feedback control with flatness‐based feedforward control
dimensional sensors for absolute motion measurement. On a larger for vibration control of a piezoelectric actuator installed on a single degree
scale simulation, Nelson, Rajamani, Gastineau, Wojtkiewicz, and of freedom (SDOF) system subjected to random excitations. Flatness‐
Schultz (2015) use feedback P‐control in a decentralized configuration based control as defined in Rigatos (2015) uses the differential flatness
for peak strain reduction of a dynamic model representation of the property of the system where all state variables can be written in terms
Cedar Avenue‐tied arch steel bridge located between Bloomington of a set of specific variables and their derivatives (so‐called flat outputs).
and Eagan, Minnesota. They use velocity feedback output measure- Li, Li, and Mo (2014) use a piezoelectric multimode control strat-
ments and six semiactive control devices (Adeli & Kim, 2009; Gutierrez egy that combines extended state observer, proportional‐derivative
Soto & Adeli, 2013b). They conclude that large structures require (PD) controller, and acceleration feedforward control for vibration
INVITED REVIEW 3 of 14

reduction of a stiff plate with all sides clamped in an experimental solutions, the addition of gradient optimization is able to obtain an
setup. Avci et al. (2015) use Proportional‐Integral (PI) feedback control understanding of global elliptical trajectories of both agents.
and acceleration feedforward control to investigate pick‐and‐place of
microobjects using two fingers of a microhand experiment.
2.6 | Model predictive control

2.3 | Optimal control Predictive control algorithms predict the change in the dependent
variables of the system caused by changes in the independent vari-
Optimal control is based on a control law that minimizes or maximizes
ables. Model predictive control (MPC) algorithms use a model to antic-
an objective function subject to constraints. The goal is to obtain opti-
ipate future evolution of the dynamic process to optimize the control
mal performance specifications for a given purpose (Ogata, 2010). This
signal. Most linear empirical models are obtained by system identifica-
is usually done off‐line (Vrabie et al., 2013).
tion. MPC has the advantages of the combined feedforward and feed-
back control. Qin and Badgwell (2003) present a brief history of the
2.4 | Classical optimal feedback control evolution of MPC algorithm starting with LQG and compare industrial
applications of MPC algorithm in linear and nonlinear systems. Chu
The preferred control algorithm with wide feedback control
et al. (2004) present a comparative study of feedback and feedforward
applications is concerned with operating a dynamic system at a mini-
MPC of nonlinear systems whereas Negenborn and Maestre (2014)
mum cost (Adeli & Saleh, 1998). Quadratic optimal control approach
provide an overview of MPC used for distributed control.
provides a systematic way to obtain a feedback gain using desired
Koerber and King (2013) use MPC to control the torque and
closed‐loop poles such that there is an acceptable trade‐off between
collective pitch of a wind turbine. Chandan and Alleyne (2013) use
acceptable response and the energy required (Ogata, 2010). These
centralized and decentralized MPC to regulate the temperature of a
approaches are the linear quadratic regulator (LQR; Adeli & Saleh,
9‐zone 3‐story square building and an 11‐zone circular building. They
1999) and a variation of the LQR with addition of a Kalman filter
compare decentralized and centralized control strategies. Because
observer (Simon, 2006) known as linear quadratic Gaussian (LQG) con-
decentralized control results in a suboptimal performance, an
trol algorithm. These control algorithms optimize a cost function to
optimality loss parameter was determined. In order to compare the
determine the feedback gain to the system. Lei, Wu, and Lin (2012)
robustness of the decentralized scheme with the centralized control,
use a decentralized LQG‐based algorithm for vibration control a 20‐
a fault propagation matrix is derived for sensing and/or
story steel building benchmark structure subjected to seismic loading.
communication faults.
They divided the structure into two and three substructures and used
Riverso, Farina, and Ferrari‐Trecate (2013) combine the MPC
recursive Kalman filter estimation to obtain interface measurements
strategy with plug‐and‐play capabilities to distribute power in a net-
between substructures.
work system composed of four power generation areas coupled
Xiao and Jing (2015) use a frequency domain method based on
through tie lines. The plug‐and‐play part occurs when a controlled sub-
Volterra series for feedback control of a vehicle suspension model.
system alternates between on‐line (plug) and off‐line (unplugged) oper-
The Volterra series method is derived from series approximation the-
ation. When a controller goes off‐line or when another subsystem is
ory and relates the nonlinear output spectrum with characteristic
added, the neighboring subsystems can adapt to this change effectively.
parameters.
Liu, Liu, Feng, and Rong (2014) compare centralized and
decentralized MPC and LQG control strategies for a chemical reactor
2.5 | Stochastic control that controls the reactor–separator process of two continuously
Stochastic control deals with the existence of uncertainty either in stirred tank reactors using two cases of gain computation considering
observations or in noise that drives the evolution of a system (Dai, (a) full state feedback and (b) only output feedback is available. For
Wang, & Zhang, 2015). Nayyar, Mahajan, and Teneketzis (2013) pro- the latter case, they report that using a Kalman filter as state observer
pose a stochastic control using Markov decision theory (Castillo guides the tuning of the decentralized MPC.
et al., 2016; Wang, Yajima, Liang, & Castaneda‐Lopez, 2015) where Yao, Jiang, Wen, Wu, and Cheng (2015) modify MPC by using a
the subsystems share partial history (past observations and control) network delay compensator to solve power stability problems and
with each other. They use two subsystems and controllers as an exam- reduce oscillations of tie lines of a wide‐area power system network.
ple and reformulate a decentralized control strategy into a centralized They compare the modified control method with the MPC without
control by using a coordinator that makes a decision based on common delay compensator using a 10‐machine 39‐bus system in New England
information shared by controllers. under uncertainty and communication delays of signal transmissions
Lin and Cassandras (2015) study stochastic decentralized control from remote areas.
of two cooperative autonomous agents (Lin, Descamps, Gaud, Hilaire Zeng and Wang (2015) study the combination of stochastic and
and Koukam, 2015; Pinto et al., 2014) in a 2‐D mission space with MPC control algorithms for energy consumption of vehicle battery
the objective of tracking elliptical trajectories and minimizing uncer- management of hybrid electric vehicles using road grade preview as
tainty. The stochastic control part is solved using an Infinitesimal Per- observer. The stochastic control part is done using Markov chains
turbation Analysis that obtains unbiased gradient estimates that can and solved using dynamic programming (Adeli & Ge, 1989). They per-
be incorporated in standard gradient‐based optimization algorithms. form 24 different trip simulations using a mountainous road terrain in
They conclude that because interconnected agents have local optimal Mariposa, California. They conclude that using the road grade preview
4 of 14 INVITED REVIEW

information as constraint input for the MPC controller improves the fuel control algorithm is verified using a 10‐story frame with an active
efficiency of the vehicle. MPC is also used for distribution of polymers in tuned mass damper.
batch processes (Corbett, Macdonald, & Mhaskar, 2015), estimation of
traffic states in large‐scale urban traffic networks (Hajiahmadi, Haddad,
De Schutter, & Geroliminis, 2015), energy efficiency of office buildings
3.2 | L2 control
(Sturzenegger, Gyalistras, Morari, & Smith, 2015), and dampening of In mathematics, Lp spaces are function spaces defined by the general-
vibration and fatigue of wind turbines (Evans, Cannon, & Kouvaritakis, ization of the p‐norm for finite dimensional vector spaces. In control,
2015; Lin, Kruger, Zhang, Wang, Lamont, & Chaar, 2015). the output signal can be defined in vector form to allow for the calcu-
lation of p‐norms to be used in the control algorithm. Cui and Jia
(2012) use L2 and L∞ for disturbance attenuation of four autonomous
3 | R O B U S T C O N T RO L agents following different directional protocols subjected to varying
time delays. Yan, Yan, Zhang, and Shi (2015) propose an event‐trig-
Robust controllers have been developed in feedback control to deal gered L2 control method for a linear‐inverted pendulum example con-
with bounded uncertainties. It is assumed that certain variables are nected in a communication network considering communication
unknown but remain bounded. The H‐infinity (H∞) method is one such delays. Hanifzadegan and Nagamune (2015) use two controllers for a
robust control method that guarantees stability of the control in an ball screw drive: (a) H∞ to track position along a moving trajectory,
uncertain dynamical system (Levine, 1996). and (b) gain‐scheduled control to reduce vibration along with L2‐gain
bound to guarantee stability.

3.1 | H2 and H∞ control


Looking at the feedback control block diagram in Figure 1(b) with some
3.3 | Sliding mode control
disturbance, the goal of the controller is to keep the effect of distur- Sliding mode control (SMC) is among the popular approaches for con-
bance on the output signal small. The disturbance rejection perfor- trol of nonlinear systems. However, chattering in SMC causes prob-
mance will depend on the size of the transfer function (relationship lems. Wang and Adeli (2012) present algorithms for reducing
between the disturbance signal and the output signal in a close loop). chattering. They propose a time‐varying method for determining the
The size of the transfer function influences the impact of the distur- sliding gain function in SMC.
bance signal on the performance variables contained in the output sig- Wang and Adeli (2015b) present a robust‐filtered SMC algorithm
nal (Levine, 1996). In order to quantify the size of the transfer function, for vibration control of high‐rise buildings subjected to wind loading.
mathematical norms such as H2 and H∞ are used. An early review and They provide a method to find the control force based on the equiva-
historical perspective of H2 and H∞ and their performance measure- lent control force principle and evaluate the performance of the algo-
ments in optimal control theory was presented by Doyle, Glover, rithm by application to a 76‐story building equipped with an active
Khargonekar, and Francis (1989). Subsequently, a review of robust tuned mass damper on the roof. They report that the new algorithm
optimal vibration control algorithms was presented by Zang et al. is more robust to structural stiffness uncertainties compared with the
(2005). LQG algorithm and another implementation of SMC.
Wang (2011) presents a time‐delayed H∞ control algorithm that
uses homotopic transformation to obtain decentralized configuration
of a five‐story shear frame structure subjected to earthquake loading.
3.4 | Backstepping control
Homotopic transformation is performed using linear matrix inequalities Backstepping is a stabilizing control technique developed by Kokotovic
constraints to transform a centralized controller configuration into (1990). It is an iterative procedure used in nonlinear systems, where
multiple uncoupled decentralized controllers. They study two the existing states of a first‐order model are used recursively along
decentralized feedback control patterns: (a) no information overlap with a Lyapunov function to stabilize the steady motions of another
between subsystems and (b) with information overlap. second‐order model (Sepulchre, Jankovic, & Kokotovic, 1997).
Fallah and Taghikhany (2013) use the time‐delay H2/LQG control He, Ge, and Huang (2015) use backstepping control and boundary
strategy to study decentralized control performance of a cable‐stayed constraints to control a moving string with varying length and speed
bridge subjected to seismic loading. They reduced the large‐scale and compare it with a PD controller. Wang and Lin (2015) propose a
structure to a simpler model using information from the dominant fre- decentralized adaptive backstepping control to track the output of
quencies of the structure. two interconnected inverted pendulums subject to unknown time‐
Based on the theories of the bounded real lemma and the linear varying parameters. Similarly, Bresch‐Pietri et al. (2014) derive a
matrix inequalities, Li and Adeli (2016) present a novel discrete‐time backstepping control algorithm with delay compensation for transport
robust H2/H∞ algorithm for vibration control of smart structures that of fuel in a gasoline engine equipped with indirect injection subjected
not only reduces the structural peak response due to external dynamic to time‐delay and varying bounded input. Fan, Liu, Shen, and Wang
forces but is also robust and stable in the presence of parametric (2014) use backstepping control to compensate the stochastic failure
uncertainties. The uncertainties of structural parameters are consid- of two actuators.
ered in the time domain as opposed to the frequency domain to facil- Xu (2015) combines SMC with MPC to obtain specific positions in
itate practical implementations. The effectiveness of the new adaptive a piezoelectric microposition and nanoposition experimental setup.
INVITED REVIEW 5 of 14

They use SMC to deal with model nonlinearities and disturbances and distributed control theory and optimization (Figure 2) can lead to
MPC to attenuate the position error. design of robust and secure multiagent‐networked systems that can
Su and Jia (2015) combine adaptive backstepping control with be applied for control of unmanned aerial vehicles, robotics, and so on.
input shaping to mitigate the elastic vibration and to track the trajec- In recent years, researchers have tried to use artificial and
tory of hypersonic vehicles (Pu, Tan, Fan, & Yi, 2014, Sun, Jing, Cheng, computational intelligence concepts such as neural networks (Cabessa
& Xu, 2015). Input shaping is a technique that identifies the modes of a & Siegelmann, 2014; Donnarumma, Prevete, Chersi, & Pezzulo, 2015),
flexible structure a priori to modify the input signal and use it as fuzzy logic (Quirós Alonso, Díaz, & Montes, 2014), evolutionary com-
feedforward control to suppress residual vibrations. They evaluate puting (Cheng, Zhang, Caraffini, & Neri, 2015; Martínez‐Ballesteros
the control method by dividing the vehicle into subsystems under Bacardit, & Riquelme, 2015), machine learning (Mesejo, Ibanez Enrique
the following constraints: parameter uncertainties, external loadings, Fernandez‐Blanco, Cedron, Pazos, & Porto‐Pazos, 2015; You, Benitez‐
and input saturation. Quiroz, & Martinez, 2014), and agent‐based modeling to develop adap-
tive or intelligent control algorithms.

4 | ADAPTIVE AND/OR INTELLIGENT


4.1 | Adaptive filters and wavelet‐based control
C O N TR O L
For a dynamic system to adapt, an adaptive filter can be included in the
Modern control theory is employed increasingly in more complicated control algorithm, for example, for estimation (Vahabi, Amirfattahi,
and varying environments where there is a high degree of uncertainty Ghassemi, & Shayegh, 2015), error reduction, noise reduction (Katicha
requiring a high level of adaptability. Intelligent control may be defined et al., 2014), correction of a particular behavior, or system identifica-
as the emulation of biological control processes for solving automation tion (Adeli & Jiang, 2006). An adaptive filter can be defined as a filter
problems (Passino, 2004). The ultimate goal of control engineering is that updates a defined transfer function according to the error signal
to implement autonomous systems that will operate independent of in order to obtain a desired output. Adaptive filters are especially use-
human actions in an unstructured and uncertain environment ful to improve a system response when there are uncertainties in a sys-
(Poznyak, Sanchez, & Yu, 2001). tem that cannot be known in advance (Hsu, 2015).
De Wolf and Holvoet (2003) propose merging three areas Kim and Adeli (2004) combine adaptive filtered‐x least mean
of research: agent computations, dynamic system theory, and square control algorithm with LQR and LQG for vibration control
decentralized control to achieve autonomous computing with the of structures subjected to strong winds and earthquakes. Ferrer,
ability to make decisions automatically, adjust to its environment Gonzalez, de Diego, and Piñero (2013) use filtered‐x adaptive filter
effectively, and use resources efficiently. Similarly, it has been to improve convergence speed and steady‐state error for active
suggested that integration of areas of computational learning, noise control of single‐channel and multi‐channel sound systems.

FIGURE 2 Integration of computational


learning, distributed control, and optimization
for development of adaptive control
algorithms
6 of 14 INVITED REVIEW

Similarly, Caiyun (2012) presents an adaptive bilinear‐filtered least Similarly, Chandra Sekhar, Kumar Sahu, Baliarsingh, and Panda (2016)
square algorithm along with Lyapunov function to guarantee the used GA and FA in combination with fuzzy PID control for a power dis-
stability of the error convergence in an active noise control system. tribution in four areas subject to load disturbances. Alfonso, Velazquez,
Hua, Cordoba‐Arenas, Warner, and Rizzoni (2015) use a nonlinear Passino, and Caicedo (2016) use honeybee social foraging for control
predictive filter to estimate states of health and charge measure- decision making in a feedback control of smart lights located in eight
ments of lithium‐ion battery packs in hybrid electric cars. The non- zones considering interzone couplings.
linear‐predictive filter processes the noise, and the process noise is Another EA variation is evolutionary game theory that uses the
also estimated as part of the solution. notion of growing populations and strategies for survival in biology
A wavelet is a wave‐type oscillation that starts at zero, raises to a or society. Game theory has gained special interests in distributed net-
certain amplitude, and ends back at zero. Wavelets are used to extract work systems for control and/or prescribing behavior. For example,
information from an unknown signal, such as an earthquake signal the replicator dynamics equation was used by Obando, Pantoja, and
(Zhou & Adeli, 2003). Adeli and Samant (2000) introduced the Quijano (2014) to solve the temperature control problem in a building
concept of wavelets for processing of traffic signals in intelligent with four rooms at different temperatures. The replicator dynamics
transportation systems. Adeli and Kim (2004) introduced the concept strategy allows resource allocation of available power to improve
of wavelets in structural control and presented a novel wavelet‐based energy usage of the overall control system.
control algorithm for vibration control of irregular high‐rise building
structures (Kim & Adeli, 2005a–c) and bridges (Kim & Adeli, 2005d).
Amini and Zabihi‐Samani (2014) present a wavelet‐based adaptive
4.3 | Agent‐based modeling
pole assignment method for vibration control of structures. Su, Liu, and
Huang (2014) present a wavelet‐based approach for identification of Wooldridge (1997) define an agent as “an encapsulated computer sys-

the instantaneous modal parameters of time‐varying systems. tem that is situated in some environment and can act flexibly and auton-
omously in that environment to meet its design objectives.” Jennings and
Bussmann (2003) propose agent‐based modeling that is appropriate
4.2 | Evolutionary computing for modeling complex systems. A complex system can take a hierarchi-
Evolutionary theory is inspired by biological evolution postulated on cal scheme and be organized as client/server, wolf/sheep, peer‐to‐
Darwinian principles. Widely used evolutionary algorithms (EAs) peer, and so on (Forero Mendoza, Vellasco, & Figueiredo, 2014).
include genetic algorithms (GAs; Lee, Yi, Lee, & Arditi, 2015; Paris, These relationships can change over time (Akar, Kara, Latifoğlu, &
Pedrino, & Nicoletti, 2015), particle swarm optimization (Iacca, Bilgiç, 2016). The designer can establish the manner in which agent
Caraffini, & Neri, 2014; Shabbir & Omenzetter, 2015; Zeng, Xu, Wu, interact with each other (Figure 3). The design decisions may form a
& Shen, 2014), artificial bee colony, and ant colony optimization communication network where one individual agent interacts with its
(Forcael et al., 2014). Another EA is the firefly algorithm (FA) inspired surrounding and be able to make a decision based on specified rules
by the mating process of fireflies that uses varying brightness of each without having the information about the entire system. Kim and
firefly's flashing lights (Chou & Pham, 2015). A survey of multiobjective Cho (2006) present a review of agent‐based modeling in control tasks
evolutionary algorithms used in engineering including control was pre- ranging from robotics to software to data mining. Leitão (2009) pre-
sented by Zhou et al. (2011). Gnana Sundari, Rajaram, and Balaraman sents a survey of multiagent distributed control in manufacturing.
(2016) use the FA to obtain optimal switch angles of 11 switches in Kilkki, Kangasrääsiö, Nikkilä, Alahäivälä, and Seilonen (2014) use
an electrical inverter. They conclude that FA achieves the harmonic agent‐based modeling for balance control in a smart grid considering
distortion reduction goal in less computational time compared with communication network between a power plant generator, energy
artificial bee colony and particle swarm optimization algorithms. demand of consumers, and a virtual power plant.

FIGURE 3 A multiagent system in canonical view with distributed hierarchical organization


INVITED REVIEW 7 of 14

Reppa, Polycarpou, and Panayiotou (2015) use agents for fault‐ Tong, Li, and Zhang (2011) present two decentralized
tolerant decentralized control and detection of multiple sensor failures backstepping control algorithms subjected to time delays to control
in a sensor network topology consisting of two interconnected robot two inverted pendulums connected by a spring that have unmeasur-
manipulators. They dedicate a local sensor within each subsystem to able states. One approach uses a neural network (NN) as state
monitor failures exclusively and isolate faulty sensors. Blanchini, observer to estimate the unmeasurable states that results in a complex
Franco, and Giordano (2015) analyze stability of decoupled subsys- network. The second approach overcomes the complexity problem by
tems using a network‐decentralized control approach where the sub- using a dynamic surface control scheme that combines a multiple slid-
systems share information only when they are connected by a node. ing surface with a low‐pass filter.
Their examples include a water distribution network with four reser- Mannava, Balakrishnan, Tang, and Landers (2012) use optimal
voirs connected by pipes that exchange water based on relative levels. tracking control with NN as an observer to estimate disturbances,
Agent‐based modeling can be based on a leader–follower strategy. parameter variations, and unmodeled dynamics of a Computer
Hu and Zheng (2014) use such a strategy for decentralized tracking Numeric Control machine experiment that tracks 2‐D diamond con-
control of a multiagent system using five agents. Similarly, Zhu and tours. Nekoukar and Erfanian (2012) use SMC and fuzzy logic to con-
Jiang (2015) use a leader–follower topology of five interconnected fol- trol the dynamics of two human joints independently for the purpose
lower agents based on independent agent events subjected to time of controlling a functional electrical stimulation that assists paraplegic
delays and using limited resources. Liu (2015) apply adaptive coopera- patients with walking. There are two objectives in this control problem:
tive and leader‐following control strategies to regulate output of four (a) the activation of muscles to alleviate the effort of the upper body
subsystems using a specified communication topology and defining and (b) the lower extremity to follow a specific trajectory. They used
each agent's control direction as unknown. Sakurama, Azuma, and two strategies for each joints, adaptive terminal SMC for one and fuzzy
Sugie (2015) use gradient flow optimization using two distributed pat- logic for the other control. Weike, Bin, and Yong (2013) use a fuzzy
tern decisions to control robots as agents to achieve a cooperative NN in combination with GA and MPC to solve a power supply prob-
consensus goal. Omidi and Mahmoodi (2015) present a multiagent lem. Tong, Sui, and Li (2014) used decentralized stochastic control with
decentralized consensus positive position feedback control for vibra- a fuzzy filter as observer that estimates both unmeasured states and
tion control of a beam with fixed supports. They placed four agents unknown control directions of three nonlinear inverted pendulums.
(actuator/sensor patches) on four nodes of the beam that collaborate Wang and Adeli (2015a) present an adaptive control algorithm for
with each other to achieve the desired vibration reduction. nonlinear vibration control of large structures subjected to dynamic
Winn and Julius (2015) propose a safety control method that uses loading based on the integration of a self‐constructing wavelet neural
simulated human trajectories to establish safety boundaries. The network developed specifically for structural system identification
method is evaluated using three numerical examples: a 2‐D planar with an adaptive fuzzy SMC approach. They use a two‐part growing
vehicle, a 2‐D hybrid rotating vehicle, and a 2‐D quadrotor with two and pruning criterion to construct the hidden layer in the neural net-
propellers. They conclude that applying crowdsourcing of human tra- work automatically. Further, they develop a fuzzy compensation con-
jectories reduces the high dimensionality of the entire system and troller to reduce the chattering phenomenon. They demonstrate the
computational time. effectiveness of the model by application to vibration control of a con-
tinuous cast‐in‐place prestressed concrete box‐girder bridge subjected
to earthquake loading. Ngo and Shin (2016) use a type 2 fuzzy basis
4.4 | Multiparadigm algorithms function and Takagi Sugeno fuzzy model to control an electrohydraulic
Jiang and Adeli (2008a) present a neuro‐genetic algorithm for nonlin- actuator under unstructured uncertainties. Takagi Sugeno fuzzy model
ear active control of irregular high‐rise buildings taking into account uses fuzzy logic for system identification of nonlinear systems.
both geometrical and material nonlinearities and coupling action Baghbani, Akbarzadeh‐T, M. Akbarzadeh, and Ghaemi, (2016) use
between lateral and torsional motions and actuator dynamics. They a type 2 fuzzy with mixed H2/H∞ bounded constraints to control the
present a dynamic fuzzy wavelet neuroemulator for predicting the consumption of energy of a three degrees of freedom prismatic‐spher-
structural response in future time steps. The neuroemulator is vali- ical‐prismatic robot experimental setup. The type 2 fuzzy strategy is
dated using two irregular three‐dimensional steel building structures used to manage uncertainties, and the mixed H2/H∞ is used to mini-
(Adeli & Jiang, 2009). Jiang and Adeli (2008b) present a neurogenetic mize the disturbance effect below a certain level and the control effort
algorithm for finding the optimal control forces. energy.

FIGURE 4 Multiobjective control for energy harvesting from vehicles in motion for road handling and ride comfort
TABLE 1 Summary of articles reviewed in this paper
8 of 14

Centralized/
Author Year Decentralized Control strategy Example Remarks
Alfonso et al. 2016 Decent Intelligent swarm optimization Light control of 8 zones Honeybee social foraging for control decision making;
interzone couplings
Baghbani et al. 2016 Cent H2, H∞, type 2 fuzzy 3DOF robot Fuzzy for uncertainty; mixed H2/H∞ for control
energy effort
Chandra Sekhar et al. 2016 Decent Firefly algorithm; fuzzy four power distribution areas Combined with fuzzy PID control; compare with GA
Gnana Sundari et al. 2016 Cent Firefly; particle swarm; artificial Pulse width modulator Optimum switch angles
bee colony (11 switches)
Takeya et al. 2016 Cent N/A Bridge structure Energy harvesting
Tang et al. 2016 Cent H2, H∞ SDOF tuned mass damper Energy harvesting
Khan and Ahmad 2016 Cent N/A Bridge oscillations Energy harvesting
Li and Adeli 2016 Cent H2 and H∞ 10‐story frame Active tuned mass damper; LMI; parameteric
uncertainties
Ngo and Shin 2016 Cent Type 2 Fuzzy; H∞ SMC Electrohydraulic actuator Unstructured uncertainties;
Anubi and Clemen 2015 Cent MPC and L2 gain no application Energy regenerative
Avci et al. 2015 Cent Feedforward and PI control Two‐finger microhand Medical robotics for cell‐assembly applications with
pick and place and vibration reduction objectives
Blanchini et al. 2015 Decent Network control, optimal LQR Water distribution, interconnected Comparison with centralized LQR or stability
control systems strategies
Corbett et al. 2015 Cent MPC Chemical production Compared with PI trajectory tracking; multiple local
model optimization
Evans et al. 2015 Cent MPC combined with robust Wind turbines Vibration reduction and fatigue analysis
control
Hajiahmadi et al. 2015 Decent MPC Urban traffic networks Hybrid perimeter and switching plans
Hanifzadegan and Nagamune 2015 Cent H∞ and L2 Ball screw driver H feedforward tracking control and L2 vibration
feedback control
He et al. 2015 Cent Backstepping control Moving string Varying lengths and speed; boundary constraints
Li, J. et al. 2015 Cent PD control Simple supported beam; Virtual electromagnetic energy harvester
maglev train
Lin and Cassandras 2015 Decent Optimal stochastic control 1‐D and 2‐D mission spaces Trajectory tracking
Lin et al. 2015 Cent MPC autoregressive exogenous (ARX) model structures Seasonal analysis and prediction; wind energy
Liu 2015 Decent Distributed adaptive control Four subsystems Nonlinear multiagent system using multiple
Nussbaum‐type functions; trajectory‐tracking
performance
Moore and Moheimani 2015 Cent P‐controller Nanoelectromechanical system Self‐sensing approach by modifying physical
components; experimental setup
Nelson et al. 2015 Decent P‐control Steel arch bridge model Modal analysis for actuator placement optimization;
velocity feedback; semiactive
Omidi and Mahmoodi 2015 Cent Positive feedback, H2, and H∞ Cantilever beam Vibration reduction and system identification;
simulated and experimental validation
INVITED REVIEW

(Continues)
TABLE 1 (Continued)

Centralized/
Author Year Decentralized Control strategy Example Remarks

Reppa et al. 2015 Decent Fault‐tolerant Two interconnected robots Methodology for detecting and isolating multiple
INVITED REVIEW

sensor faults
Sakurama et al. 2015 Decent Gradient flow approach Robots Multiagent system; objective to coordinate consensus,
formation and pattern decision
Sturzenegger et al. 2015 Cent MPC Office building Cost benefit analysis; energy consumption
Su and Jia 2015 Decent Adaptive backsteeping control; Flexible hypersonic vehicle Control objectives are adaptive flight trajectory
feedforward input shape tracking and elastic vibration mitigation
Sun C. et al. 2015 Decent MPC Hybrid electric vehicles Velocity predictors
Wang and Lin 2015 Decent Adaptive backstepping Two interconnected inverted Unknown time‐varying parameters; output tracking
pendulums
Winn and Julius 2015 Cent Safety control synthesis 2‐D planar vehicle, a 2‐D hybrid Human trajectories to reduce high dimensionality of a
rotating vehicle, and a 2‐D complex system
quadrotor with two propellers.
Xiao and Jing 2015 Cent Optimal nonconvex control Vehicle suspension Frequency domain analysis, nonlinear characteristic
output spectrum concept
Xu 2015 Cent MPC Piezoelectric nanopositioning Tracking error, hysteresis model
Yan et al. 2015 Decent L2 Inverted pendulum cart Event‐triggered
Yao et al. 2015 Decent MPC Network urban traffic Wide‐area damping for power systems
Zeng and Wang 2015 Decent MPC Hybrid electric vehicles Road grade preview
Zhou and Zuo 2015 Cent H2 Simple cantilever beam Integrates energy harvesting
Zhu and Jiang 2015 Decent Event‐based Multiagent system Limiting onboard resources; leader–follower
distributed topology
Wang and Adeli (a) 2015 Cent Fuzzy SMC wavelet NN Concrete bridge Fuzzy control compensation for chattering
Wang and Adeli (b) 2015 Cent SMC; LQG 76‐story structure Active tuned mass damper
Bresch‐Pietri et al. 2014 Cent Backstepping control Gasoline engine Transport phenomena in process industry; time‐delay
compensator
Fan et al. 2014 Cent Backstepping control Two‐order nonlinear system Stochastic actuator failure
Hu and Zheng 2014 Decent Multiagent system and/or Simulation: one leader and four Leader–follower distributed scheme
adaptive tracking followers network
Kilkki et al. 2014 Decent Frequency control Smart grid Agent‐based model using four cases
Li, P. et al. 2014 Cent Positive position feedback 1‐D beam Create redundancy by using time synchronization to
aid sensor failures
Li, S. et al. 2014 Cent Multimode control strategy 2‐D plate with clamped sides Feedback and feedforward control; acceleration
compensator using a chaotic sequence
Liu et al. 2014 Decent LQG/MPC Chemical reactor Full state feedback and only output feedback
available; Kalman filter observer
Negenborn and Maestre 2014 Decent MPC No application Categorized 35 distinct approaches

(Continues)
9 of 14
10 of 14

TABLE 1 (Continued)

Centralized/
Author Year Decentralized Control strategy Example Remarks

Obando et al. 2014 Decent Population replicator dynamics four rooms Different temperature setpoints at each room,
comparison with LQR
Pu et al. 2014 Cent Robust trajectory linearization Hypersonic vehicles Uncertainty analysis and adaptive flight trajectory
control
Qu et al. 2014 Decent H∞ Six‐story numerical simulation, Double homotopy using bilinear matrix inequality
two‐story frame experiment
Tong et al. 2014 Decent Stochastic control with fuzzy Nonlinear inverted pendulum Division of subsystems: two subsystems
filter
Chandan and Alleyne 2013 Decent MPC 9‐zone 3‐story building, 11‐zone Temperature control
circular building
Fallah and Taghikhany 2013 Decent H2/LQG controller Cable‐stayed bridge Time‐delay
Ferrer et al. 2013 Cent Adaptive filter Acoustic signals Active noise control–convex combination filter
Lei et al. 2013 Decent LQR control 20‐story benchmark steel, Division of subsystems: two subsystems (Ex1), three
35‐story shear building subsystems (Ex2)
Nayyar et al. 2013 Decent Stochastic Static team of two controllers Observers and controllers share history with each
other
Riverso et al. 2013 Decent MPC, plug‐and‐play Power networks Adapt of neighboring subsystems to off‐line
controllers or addition of subsystems to an
exiting network
Rodriguez Fortun et al. 2013 Cent Integrates sky‐hook control with SDOF metrological devices Considers hysteresis and rate dependency of
flatness‐based feedforward piezoelectric actuator
control
Weike et al. 2013 Cent Fuzzy NN; GA; predictive control Power plant Compensate for process disturbances, measurement
noise and modeling errors
Cui and Jia 2012 Decent L2‐ L∞ control four+ agents multiagent system; time‐varying delays; switching
topologies; disturbance attenuation
Nekoukar and Erfanian 2012 Decent Sliding mode control, fuzzy logic Two cantilever beams resembling Trajectory tracking of upper and lower extremity joint
joint and two muscles dynamics
Mannava et al. 2012 Cent NN Computer machine control Tracking control; diamond contour, feedforward and
feedback control
Tong et al. 2011 Decent Adaptive NN and backstepping Two inverted pendulums Consider unmeasurable states and time delays
Wang 2011 Decent H∞ five‐story shear frame Homotopic transformation to obtain decentralized
configuration; time delayed

Note. 1‐D = one‐dimensional, 2‐D = two‐dimensional, DOF = degree of freedom, GA = genetic algorithms, LMI = linear matrix inequalities, LQG = linear quadratic Gaussian, LQR = linear quadratic regulator, MPC = model
predictive control, NN = neural network, SMC = sliding mode control, PD = proportional‐derivative, PID = proportional‐integral‐derivative.
INVITED REVIEW
INVITED REVIEW 11 of 14

5 | CONTROL AND ENERGY HARVESTING multiobjective control and energy‐harvesting systems is another
promising multidisciplinary research area in the years to come.
Sustainability as defined by conserving energy or using alternative
energy sources has been an important research trend in recent RE FE RE NC ES
years. Some investigators have explored the idea of using the Adeli, H., & Ge, Y. (1989). A dynamic programming method for analysis of
bridges under multiple moving loads. International Journal for Numerical
mechanical vibrations of structures as input to harvest usable elec-
Methods in Engineering, 28, 1265–1282.
trical power. Research into simultaneous control and energy har-
Adeli, H., & Jiang, X. (2006). Dynamic fuzzy wavelet neural network model
vesting presents complexities that have motivated many for structural system identification. Journal of Structural Engineering,
investigations. Wang and Inman (2012) present a survey of control ASCE, 132(1), 102–111.
strategies for simultaneous vibration suppression and energy har- Adeli, H., & Jiang, X. (2009). Intelligent infrastructure—Neural networks,
vesting using piezoelectric materials. Peigney and Siegert (2013) wavelets, and chaos theory for intelligent transportation systems and
smart structures. Taylor & Francis, Boca Raton, Florida: CRC Press.
use a piezoelectric device to harvest energy in traffic‐induced vibra-
Adeli, H., & Kim, H. (2004). Wavelet‐hybrid feedback least mean square
tions in a prestressed concrete highway bridge. Khan and Admad
algorithm for robust control of structures. Journal of Structural Engineer-
(2016) present a review of various devices used for energy harvest- ing, 130(1), 128–137.
ing of bridge structure vibrations.
Adeli, H., & Kim, H. (2009). Wavelet‐based vibration control of smart buildings
Zhou and Zuo (2015) investigate a self‐powered H2‐controlled and bridges. Taylor & Francis, Boca Raton, Florida: CRC Press.
actuator installed on a flexible cantilever beam for simultaneous vibra- Adeli, H., & Saleh, A. (1998). Integrated structural/control optimization of
tion control and energy harvesting. Their results show that the accu- large adaptive/smart structures. International Journal of Solids and Struc-
tures, 35(28–29), 3815–3830.
mulated harvested energy is higher than the consumed energy used
by the controller. Anubi and Clemen (2015) study self‐powered MPC Adeli, H., & Saleh, A. (1999). Control, optimization, and smart structures—
High‐performance bridges and buildings of the future. New York, USA:
of an SDOF simulation under stability constraints and use L2‐norm to
John Wiley and Sons.
minimize the estimation error. Li et al. (2015) investigate the harvest
Adeli, H., & Samant, A. (2000). An adaptive conjugate gradient neural net-
energy and PD vibration control using an electromagnetic device work—Wavelet model for traffic incident detection. Computer‐Aided
installed on a platform whereas a full‐scale maglev train travels across. Civil and Infrastructure Engineering, 15(4), 251–260.
Takeya, Sasaki, and Kobayashi (2016) replace the damper of a Akar, S. A., Kara, S., Latifoğlu, F., & Bilgiç, V. (2016). Analysis of the com-
dual‐mass tuned mass damper with an electromagnetic transducer to plexity measures in the EEG of schizophrenia patients. International
Journal of Neural Systems, 26(2) 1650008 (13 pages).
harvest energy from low‐frequency mechanical vibrations of a bridge.
Alfonso, W., Velazquez, J. J., Passino, K. M., & Caicedo, E. F. (2016). A
Tang, Liu, Cui, and Zuo (2016) use H2 and H∞ norms for optimal tuning
honeybee social foraging algorithm for feedback control of smart lights.
of an SDOF tuned mass damper (Gutierrez Soto, & Adeli, 2013a) con- Engineering Applications of Artificial Intelligence, 48, 13–31.
nected to an R‐L‐C (resistor‐inductor‐capacitor) circuit for simulta- Amini, F., & Zabihi‐Samani, M. (2014). A wavelet‐based adaptive pole
neous vibration control and energy harvesting under random assignment method for structural control. Computer‐Aided Civil and
excitations (Cassidy, Scruggs, Behrens, & Gavin, 2011). Figure 4 Infrastructure Engineering, 29(6), 464–477.

depicts an example of sustainable control using energy harvesting Anubi, O. M., & Clemen, L. (2015). Energy‐regenerative model predictive
control. Journal of the Franklin Institute, 352(5), 2152–2170.
and vibration reduction from vehicles and civil infrastructure.
Avci, E., Ohara, K., Nguyen, C.‐N., Theeravithayangkura, C., Kojima, M.,
Tanikawa, T., & Arai, T. (2015). High‐speed automated manipulation
of microobjects using a two‐fingered microhand. Industrial Electronics,
6 | C O N CL U S I O N IEEE Transactions on, 62(2), 1070–1079.
Baghbani, F., Akbarzadeh‐T, M. R., Akbarzadeh, A., & Ghaemi, M. (2016).
A summary of the articles reviewed in this paper is presented in Robust adaptive mixed H2/H∞ interval type‐2 fuzzy control of
nonlinear uncertain systems with minimal control effort. Engineering
Table 1 in reverse chronological order. There has been an increasing
Applications of Artificial Intelligence, 46, 88–102.
interest in the combination of feedforward and feedback control
Blanchini, F., Franco, E., & Giordano, G. (2015). Networked‐decentralized
strategies. Novel autonomous control systems intersect distributed control strategies for stabilization. Automatic Control, Transactions on,
control theory, agent‐based modeling, bioinspired computational 60(2), 491–497.
learning, and optimization algorithms. Bresch‐Pietri, D., Chauvin, J., & Petit, N. (2014). Prediction‐based stabiliza-
The authors are currently developing novel adaptive control algo- tion of linear systems subject to input‐dependent input delay of
integral‐type. Automatic Control, IEEE Transactions on, 59(9), 2385–
rithms to simultaneously perform vibration suppression and energy
2399.
minimization through integration of the agent technology and
Cabessa, J., & Siegelmann, H. T. (2014). The super‐turing computational
replicator dynamics from the evolutionary game theory. The use of power of evolving recurrent neural networks. International Journal of
agents and a decentralized approach enhances the robustness of the Neural Systems, 24(8), 1450029 (22 pages).
entire vibration control system. Another line of research would be Caiyun, W. (2012). Active Noise Control Based On Adaptive Bilinear F‐L
integration of two fields of vibration control and structural health Algorithm. In Computer distributed control and intelligent environmental
monitoring (CDCIEM), International Conference on, 5–6. (pp. 89–92).
monitoring (Karami & Akbarabadi, 2016).
Cassidy, I., Scruggs, J. T., Behrens, S., & Gavin, H. P. (2011). Design and
Vibration energy‐harvesting systems can be the source of energy
experimental characterization of an electromagnetic transducer for
for active and semiactive control and health‐monitoring systems to large‐scale vibratory energy harvesting applications. Journal of Intelli-
make them self‐powered and sustainable. Integration of a gent Materials Systems and Structures, 22(17), 2009–2024.
12 of 14 INVITED REVIEW

Castillo, A., Calviño, A., Grande, Z., Sánchez‐Cambronero, S., Gallego, I., Gnana Sundari, M., Rajaram, M., & Balaraman, S. (2016). Application of
Rivas, A., & Menéndez, J. A. (2016). A Markovian‐Bayesian network improved firefly algorithm for programmed PWM in multilevel inverter
for risk analysis of high speed and conventional railway lines integrating with adjustable DC sources. Applied Soft Computing, 41, 169–179.
human errors. Computer‐Aided Civil and Infrastructure Engineering, 31(3), Gutierrez Soto, M., & Adeli, H. (2013a). Tuned mass dampers. Archives of
193–218. Computational Methods in Engineering, 20(4), 419–431.
Chandan, V., & Alleyne, A. (2013). Optimal partitioning for decentralized Gutierrez Soto, M., & Adeli, H. (2013b). Placement of control devices for
thermal control of buildings. IEEE Transactions on Control Systems passive, semi‐active, and active, vibration control of structures. Scientia
Technology, 21(5), 1756–1770. Iranica ‐ Transaction A: Civil Engineering, 20(6), 1567–1578.
Chandra Sekhar, G. T., Kumar Sahu, R., Baliarsingh, A. K., & Panda, S. (2016). Hajiahmadi, M., Haddad, J., De Schutter, B., & Geroliminis, N. (2015). Opti-
Load frequency control of power system under deregulated environ- mal hybrid perimeter and switching plans control for urban traffic
ment using optimal firefly algorithm. International Journal of Electrical networks. Control Systems Technology, IEEE Transactions on, 23(2),
Power & Energy Systems, 74, 195–211. 464–478.
Cheng, J., Zhang, G., Caraffini, F., & Neri, F. (2015). Multicriteria adaptive Hanifzadegan, M., & Nagamune, R. (2015). Tracking and structural vibration
differential evolution for global numerical optimization. Integrated control of flexible ball‐screw drives with dynamic variations.
Computer‐Aided Engineering, 22(2), 103–117. Mechatronics, IEEE/ASME Transactions on, 20(1), 133–142.
Chou, J. S., & Pham, A. D. (2015). Smart artificial firefly colony‐based sup- Hassani, V., Tjahjowidodo, T., & Nho Do, T. (2014). A survey on hysteresis
port vector regression for enhanced forecasting in civil engineering. modeling, identification and control. Mechanical Systems and Signal Pro-
Computer‐Aided Civil and Infrastructure Engineering, 30(9), 715–732. cessing, 49(1–2), 209–233. doi:10.1016/j.ymssp.2014.04.012
Chu, J.‐Z., Jang, S.‐S., & Chen, Y.‐N. (2004). A comparative study of com- He, W., Ge, S., & Huang, D. (2015). Modeling and vibration control for a
bined feedforwards/feedback model predictive control for nonlinear nonlinear moving string with output constraint. Mechatronics, IEEE/
systems. The Canadian Journal of Chemical Engineering, 82(6), 1263– ASME Transactions on, 20(4), 1886–1897.
1272. doi:10.1002/cjce.5450820613 Hsu, W. Y. (2015). Assembling a multi‐feature EEG classifier for left‐
Corbett, B., Macdonald, B., & Mhaskar, P. (2015). Model predictive quality right motor data using wavelet‐based fuzzy approximate entropy
control of polymethyl methacrylate. Control Systems Technology, IEEE for improved accuracy. International Journal of Neural Systems, 25(8)
Transactions on, 23(2), 687–692. (13 pages).
Cui, Y., & Jia, Y. (2012). L2‐L∞ consensus control for high‐order multi‐agent Hu, J., & Zheng, W. X. (2014). Adaptive tracking control of leader‐follower
systems with switching topologies and time‐varying delays. IET Control systems with unknown dynamics and partial measurements.
Theory and Applications, 6(12), 1933–1940. Automatica, 50(5), 1416–1423.

Dai, H., Wang, W., & Zhang, H. (2015). A multiwavelet neural network‐ Hua, Y., Cordoba‐Arenas, A., Warner, N., & Rizzoni, G. (2015). A multi‐scale
based response surface method for structural reliability Analysis. state‐of‐charge and state‐of‐health estimation framework using a non-
Computer‐Aided Civil and Infrastructure Engineering, 3(2), 151–162. linear predictive filter for lithium‐ion battery pack with passive balance
control. Journal of Power Systems, 280, 293–312.
De Wolf, T., Holvoet, T. (2003) “Towards autonomic computing: Agent‐based
modelling, dynamical systems analysis, and decentralised control”. Iacca, G., Caraffini, F., & Neri, F. (2014). Multi‐strategy coevolving aging
Proceedings IEEE International Conference on Industrial Informatics. particle optimization. International Journal of Neural Systems, 24(1)
1450008 (19 pages).
Donnarumma, F., Prevete, R., Chersi, F., & Pezzulo, G. (2015). A program-
Jennings, N., & Bussmann, S. (2003). Agent‐based control systems: Why
mer‐interpreter neural network architecture for prefrontal cognitive
are they suited to engineering complex systems? Control Systems. IEEE
control. International Journal of Neural Systems, 25(6) 1550017
Transactions on, 23(3), 61–73.
(16pages).
Jiang, X., & Adeli, H. (2008a). Dynamic fuzzy wavelet neuroemulator for
Dorf, R. C., & Bishop, R. H. (2008). Modern control systems. Upper Saddle
nonlinear control of irregular high rise building structures. International
River, NJ, USA: Prentice Hall.
Journal for Numerical Methods in Engineering, 74(7), 1045–1066.
Dorf, R. C., & Kusiak, A. (1994). Handbook of manufacturing and automation.
Jiang, X., & Adeli, H. (2008b). Neuro‐genetic algorithm for nonlinear active
New York, NY, USA: John Wiley & Sons.
control of high rise buildings. International Journal for Numerical Methods
Doyle, J. C., Glover, K., Khargonekar, P. P., & Francis, B. A. (1989). State‐ in Engineering, 75(8), 770–786.
space solutions to standard H2 and H∞ control problems. Automatic
Karami, K., Akbarabadi, S., 2016. Develop smart structure using damage
Control, IEEE Transactions on, 34(8), 831–847.
detection and vibration control, computer‐aided civil and infrastructure
Evans, M., Cannon, M., & Kouvaritakis, B. (2015). Robust MPC tower engineering, to appear.
damping for variable speed wind turbines. Control Systems Technology, Katicha, S. W., Flintsch, G., & Bryce, J. (2014). Wavelet Denoising of TSD
IEEE Transactions on, 23(1), 290–296. Deflection Slope Measurements for Improved Pavement Structural
Fallah, A. Y., & Taghikhany, T. (2013). Time‐delayed decentralized H2/LQG Evaluation. Computer‐Aided Civil and Infrastructure Engineering, 29(6),
controller for cable‐stayed bridge under seismic loading. Structural 399–415.
Control and Health Monitoring, 20(3), 354–372. Khan, F. U., & Ahmad, I. (2016). Review of energy harvesters utilizing bridge
Fan, H., Liu, B., Shen, Y., & Wang, W. (2014). Adaptive failure compensation vibrations. Shock and Vibration, 2016, (21 pages).1340402
control for uncertain systems with stochastic actuator failures. Kilkki, O., Kangasrääsiö, A., Nikkilä, A., Alahäivälä, A., & Seilonen, I. (2014).
Automatic Control, IEEE Transactions on, 59(3), 808–814. Agent‐based modeling and simulation of a smart grid: A case study of
Ferrer, M., Gonzalez, A., de Diego, M., & Piñero, G. (2013). Convex combi- communication effects on frequency control. Engineering Applications
nation filtered‐x algorithms for active noise control systems. IEEE of Artificial Intelligence., 33, 91–98.
Transactions on audio, speech, and language processing, 21(1), 156–167. Kim, H., & Adeli, H. (2004). Hybrid feedback‐LMS algorithm for structural
Forcael, E., González, V., Orozco, F., Vargas, S., Moscoso, P., & Pantoja, A. control. Journal of Structural Engineering, ASCE., 130(1), 120–127.
(2014). Ant colony optimization model for tsunamis evacuation routes. Kim, H., & Adeli, H. (2005a). Hybrid control of smart structures using a
Computer‐Aided Civil and Infrastructure Engineering, 29(10), 723–737. novel wavelet‐based algorithm. Computer‐Aided Civil and Infrastructure
Forero Mendoza, L., Vellasco, M., & Figueiredo, K. (2014). Intelligent Engineering, 20(1), 7–22.
multiagent coordination based on reinforcement hierarchical neuro‐ Kim, H., & Adeli, H. (2005b). Hybrid control of irregular steel highrise build-
fuzzy models. International Journal of Neural Systems., 24(8) 1450031 ing structures under seismic excitations. International Journal for
(20 pages). Numerical Methods in Engineering, 63(12), 1757–1774.
INVITED REVIEW 13 of 14

Kim, H., & Adeli, H. (2005c). Wind‐induced motion control of 76‐story Moore, S., & Moheimani, S. O. (2015). Vibration control with MEMS elec-
benchmark building using the hybrid damper tuned liquid column trostatic drives: A self‐sensing approach. Control Systems Technology,
damper system. Journal of Structural Engineering, ASCE, 131(12), IEEE Transactions on, 23(3), 1237–1244.
1794–1802. Nayyar, A., Mahajan, A., & Teneketzis, D. (2013). Decentralized stochastic
Kim, H., & Adeli, H. (2005d). Wavelet hybrid feedback‐LMS algorithm for control with partial history sharing: A common information approach.
robust control of cable‐stayed bridges. Journal of Bridge Engineering, IEEE Transactions on Automatic Control, 58(7), 1644–1658.
ASCE, 10(2), 116–123. Negenborn, R., & Maestre, J. (2014). Distributed model predictive control:
Kim, K.‐J., & Cho, S.‐B. (2006). A comprehensive overview of the applica- An overview and roadmap of future research opportunities. Control Sys-
tion of artificial life. MIT Press, Artificial Life, 12(1), 153–182. tems Technology, IEEE Transactions on, 34(4), 87–97.
Koerber, A., & King, R. (2013). Combined Feedback‐Feedforward Control Nekoukar, V., & Erfanian, A. (2012). A decentralized modular control frame-
of Wind Turbines Using State‐Constrained Model Predictive Control. work for robust control of FES‐activated walker‐assisted paraplegic
Control Systems Technology, IEEE Transactions on, 21(4), 1117–1128. walking using terminal sliding mode and fuzzy logic control. Biomedical
Engineering, IEEE Transactions on., 59(10), 2818–2827.
Kokotovic, P. V. (1990). The joy of feedback: Nonlinear and adaptive. Con-
trol Systems Technology, IEEE Transactions on, 12(3), 7–17. Nelson, G., Rajamani, R., Gastineau, A., Wojtkiewicz, S., & Schultz, A.
(2015). Bridge life extension using semiactive vibration control.
Lee, H. G., Yi, C. Y., Lee, D. E., & Arditi, D. (2015). An advanced stochastic
Mechatronics, IEEE/ASME Transactions on, 20(1), 207–216.
time‐cost tradeoff analysis based on a CPM‐guided multi‐objective
genetic algorithm. Computer‐Aided Civil and Infrastructure Engineering., Ngo, P. D., & Shin, Y. C. (2016). Modeling of unstructured uncertainties and
30(10), 824–842. robust controlling of nonlinear dynamic systems based on type‐2 fuzzy
basis function networks. Engineering Applications of Artificial Intelligence,
Lei, Y., Wu, D. T., & Lin, Y. (2012). A decentralized control algorithm for
53, 74–85.
large scale systems. Computer‐Aided Civil and Infrastructure Engineering.,
27(1), 2–13. Obando, G., Pantoja, A., & Quijano, N. (2014). Building temperature control
based on population dynamics. IEEE Transactions on Control Systems
Leitão, P. (2009). Agent‐based distributed manufacturing control: A state‐
Technology, 22(1), 404–412.
of‐the‐art survey. Engineering Applications of Artificial Intelligence., 22
(7), 979–991. Ogata, K. (2010). Modern Control Engineering (Fifth ed.). Upper Saddle River,
NJ, USA: Prentice Hall.
Levine, W. S. (1996). The control handbook. Boca Raton: Florida. CRC Press.
Omidi, E., & Mahmoodi, N. (2015). Hybrid positive feedback control for
Li, Z., Adeli, H., 2016. New discrete‐time robust H2/H∞ algorithm for vibration
active vibration attenuation of flexible structures. Mechatronics, IEEE/
control of smart structures using linear matrix inequalities. Engineering
ASME Transactions on, 20(4), 1790–1797.
Applications of Artificial Intelligence, in press.
Paris, P. C. D., Pedrino, E. C., & Nicoletti, M. C. (2015). Automatic learning
Li, P., Li, L., Song, G., & Yu, Y. (2014). Wireless sensing and vibration control
of image filters using cartesian genetic programming. Integrated Com-
with increased redundancy and robustness design cybernetics. IEEE
puter‐Aided Engineering, 22(2), 135–151.
Transactions on, 44(11), 2076–2087.
Passino, K. (2004). Biomimicry for optimization, control, and automation. UK:
Li, S., Li, J., & Mo, Y. (2014). Piezoelectric multimode vibration control for
Springer‐Verlag London.
stiffened plate using ADRC‐based acceleration compensation. Industrial
Electronics, IEEE Transactions on, 61(12), 6892–6902. Peigney, M., & Siegert, D. (2013). Piezoelectric energy harvesting from
traffic‐induced bridge vibrations. Smart Materials and Structures., 22
Li, J., Li, J., Zhou, D., Cui, P., Wang, L., & Yu, P. (2015). The active control of
(9), 1–11.
maglev stationary self‐excited vibration with a virtual energy harvester.
Industrial Electronics, IEEE Transactions on, 62(5), 2942–2951. Pinto, T., Vale, Z., Sousa, T. M., Praça, I., Santos, G., & Morais, H. (2014).
Adaptive learning in agents behaviour: A framework for electricity mar-
Lin, X., & Cassandras, C. (2015). An optimal control approach to the multi‐
kets simulation. Integrated Computer‐Aided Engineering, 21(4), 399–415.
agent persistent monitoring problem in two‐dimensional spaces. Auto-
matic Control, IEEE Transactions on, 60(6), 1659–1664. Poznyak, A. S., Sanchez, E. N., & Yu, W. (2001). Differential neural networks
for robust nonlinear control. Singapore: World Scientific.
Lin, Y., Kruger, U., Zhang, J., Wang, Q., Lamont, L., & Chaar, L. (2015). Sea-
sonal analysis and prediction of wind energy using random forests and Pu, Z., Tan, X., Fan, G., & Yi, J. (2014). Uncertainty analysis and robust tra-
ARX model structures. Control Systems Technology, IEEE Transactions jectory linearization control of a flexible air‐breathing hypersonic
on., 23(5), 1994–2002. vehicle. Acta Astronautica, 101, 16–32.

Lin, Y., Descamps, P., Gaud, N., Hilaire, V., & Koukam, A. (2015). Multi‐ Qin, S. J., & Badgwell, T. A. (2003). A survey of industrial model predictive
agent system for intelligent scrum project management. Integrated control technology. Control Engineering Practice, 11, 733–764.
Computer‐Aided Engineering, 22(3), 281–296. Quirós, P., Alonso, P., Díaz, I., & Montes, S. (2014). On the use of fuzzy par-
Liu, L. (2015). Adaptive cooperative output regulation for a class of nonlin- titions to protect data. Integrated Computer‐Aided Engineering, 21(4),
ear multi‐agent systems. Automatic Control, IEEE Transactions on, 60(6), 355–366.
1677–1682. Reppa, V., Polycarpou, M., & Panayiotou, C. (2015). Decentralized isolation
Liu, S., Liu, J., Feng, Y., & Rong, G. (2014). Performance assessment of of multiple sensor faults in large‐scale interconnected nonlinear sys-
decentralized control systems: An iterative approach. Control Engineer- tems. Automatic Control, IEEE Transactions on, 60(6), 1582–1596.
ing Practice, 22, 252–263. Rigatos, G. G. (2015). Nonlinear control and filtering using differential flatness
Mannava, A., Balakrishnan, S. N., Tang, L., & Landers, R. G. (2012). Optimal approaches: Applications to electromechanical systems. Switzerland:
tracking control of motion systems. Control Systems Technology, IEEE Springer International Publishing.
Transactions on, 20(6), 1548–1558. Riverso, S., Farina, M., & Ferrari‐Trecate, G. (2013). Plug‐and‐play
Martínez‐Ballesteros, M., Bacardit, J., & Riquelme, J. C. (2015). Enhancing decentralized model predictive control for linear systems. IEEE Transac-
the scalability of evolutionary algorithms to discover quantitative asso- tions on Automatic Control, 58(10), 2608–2614.
ciation rules in large‐scale datasets. Integrated Computer‐Aided Rodriguez Fortun, J. M., Orus, J., Alfonso, J., Gimeno, F. B., & Castellanos, J.
Engineering., 22(1), 21–39. A. (2013). Flatness‐based active vibration control for piezoelectric actu-
Mesejo, P., Ibanez Enrique Fernandez‐Blanco, O., Cedron, F., Pazos, A., & ators. IEEE/ASME Transactions on Mechatronics, 18(1), 221–229.
Porto‐Pazos, A. B. (2015). Artificial neuron‐glia networks learning para- Sakurama, K., Azuma, S.‐I., & Sugie, T. (2015). Distributed controllers for
digm based on cooperative coevolution. International Journal of Neural multi‐agent coordination via gradient‐flow approach. Automatic Control,
Systems., 25(4) 1550012 (19 pages). IEEE Transactions on, 60(6), 1471–1485.
14 of 14 INVITED REVIEW

Sepulchre, R., Jankovic, M., & Kokotovic, P. V. (1997). Constructive nonlinear Wang, H., Yajima, A., Liang, R. Y., & Castaneda‐Lopez, H. (2015). Bayesian
control. New York, USA: Springer. modeling of external corrosion in underground pipelines based on the
Shabbir, F., & Omenzetter, P. (2015). Particle swarm optimization with integration of Markov chain Monte Carlo techniques and clustered
sequential niche technique for dynamic finite element model updating. inspection data. Computer‐Aided Civil and Infrastructure Engineering, 30
Computer‐Aided Civil and Infrastructure Engineering, 30(5), 359–375. (4), 300–316.

Simon, D. (2006). Optimal state estimation: Kalman, H∞, and nonlinear Weike, Y., Bin, L., Yong, X., 2013. Fuzzy neural networks and GA based pre-
approaches. Hoboken, NJ, USA: John Wiley & Sons, Inc. dictive control for active power filter. Proceedings of the fifth
conference on measuring technology and mechatronics automation.
Sturzenegger, D., Gyalistras, D., Morari, M., & Smith, R. (2015). Model pre-
598–601
dictive climate control of a swiss office building: Implementation,
results, and cost–benefit analysis. Control Systems Technology, IEEE Winn, A., & Julius, A. (2015). Safety controller synthesis using human gen-
Transactions on, 24(1), 1–12. erated trajectories. Automatic Control, IEEE Transactions on, 60(6),
1597–1610.
Su, W. S., Liu, C. Y., & Huang, C. S. (2014). Identification of Instantaneous
modal parameters of time varying systems via a wavelet‐based Wooldridge, M., 1997. Agent‐based software engineering. Proceedings
approach and its applications. Computer‐Aided Civil and Infrastructure Institute of Electrical Engineering, 144: 26–37
Engineering, 29(4), 279–298. Xiao, Z., & Jing, X. (2015). Frequency‐domain analysis and design of linear
Su, X., & Jia, Y. (2015). Constrained adaptive tracking command shaped feedback of nonlinear systems and applications in vehicle suspensions.
vibration control of flexible hypersonic vehicles. IET Control Theory Mechatronics, IEEE/ASME Transactions on, 21(1), 506–517.
and Applications, 9(12), 1857–1868. doi:10.1049/iet-cta.2014.0750 Xu, Q. (2015). Digital sliding mode prediction control of piezoelectric
Sun, C., Hu, X., Moura, S., & Sun, F. (2015). Velocity predictors for predic- micro/nanopositioning system. Control Systems Technology, IEEE
tive energy management in hybrid electric vehicles. Control Systems Transactions on, 23(1), 297–304.
Technology, IEEE Transactions on, 23(3), 1197–1204. Yan, H., Yan, S., Zhang, H., & Shi, H. (2015). L2 control design of event‐trig-
Sun, X., Jing, X., Cheng, L., & Xu, J. (2015). A 3‐D quasi‐zero‐stiffness‐based gered networked control systems with quantizations. Journal of the
sensor system for absolute motion measurement and application in Franklin Institute., 352(1), 332–345.
active vibration control. Mechatronics, IEEE/ASME Transactions on, 20 Yao, W., Jiang, L., Wen, J., Wu, Q., & Cheng, S. (2015). Wide‐area damping
(1), 254–262. controller for power system interarea oscillations: A networked predic-
Takeya, K., Sasaki, E., & Kobayashi, Y. (2016). Design and parametric study tive control approach. Control Systems Technology, IEEE Transactions on,
on energy harvesting from bridge vibration using tuned dual‐mass 23(1), 27–36.
damper systems. Journal of Sound and Vibration, 361, 50–65. You, D., Benitez‐Quiroz, C. F., & Martinez, A. M. (2014). Multiobjective
Tang, X., Liu, Y., Cui, W., & Zuo, L. (2016). Analytical solutions to H2 and H∞ optimization for model selection in kernel methods in regression. Neural
optimizations for resonant shunted electromagnetic tuned mass Networks and Learning Systems, IEEE Transactions on, 25(10),
damper and vibration energy harvesting. Journal of Vibration and Acous- 1879–1893.
tics, 138(1), 011018. Zang, C., Friswell, M. I., & Mottershead, J. E. (2005). A review of robust
Tong, S.‐C., Li, Y.‐M., & Zhang, H.‐G. (2011). Adaptive neural network optimal design and its application in dynamics. Computers and Struc-
decentralized backstepping output‐feedback control of nonlinear tures, 83(4–5), 315–326. doi:10.1016/j.compstruc.2004.10.007
large‐scale systems with time delays. IEEE Transactions on Neural Net- Zeng, X., & Wang, J. (2015). A parallel hybrid electric vehicle energy man-
works, 22(7), 1073–1086. agement strategy using stochastic model predictive control with road
Tong, S., Sui, S., & Li, Y. (2014). Adaptive fuzzy decentralized output stabi- grade preview. Control Systems Technology, IEEE Transactions on, 23(6),
lization for stochastic nonlinear large‐scale systems with unknown 2416–2423.
control directions. IEEE Transactions on Fuzzy Systems, 22(5), Zeng, Z., Xu, J., Wu, S., & Shen, M. (2014). Antithetic method‐based particle
1365–1372. swarm optimization for a queuing network problem with fuzzy data in
Vahabi, Z., Amirfattahi, R., Ghassemi, F., & Shayegh, F. (2015). Online epi- concrete transportation systems. Computer‐Aided Civil and Infrastruc-
leptic seizure prediction using wavelet‐based bi‐phase correlation of ture Engineering., 29(10), 771–800.
electrical signal tomography. International Journal of Neural Systems, Zhou, Z., & Adeli, H. (2003). Time‐frequency signal analysis of earthquake
25(6) 1550028 (22 pages). records using Mexican hat wavelets. Computer‐Aided Civil and Infra-
Vrabie, D., Vamvoudakis, K. G., & Lewis, F. L. (2013). Optimal adaptive con- structure Engineering., 18(5), 379–389.
trol and differential games by reinforcement learning principles, The Zhou, W., & Zuo, L. (2015). A self‐powered piezoelectric vibration control
Institute of Engineering Technology. London: U.K. system with switch precharged inductor (SPCI) method. Mechatronics,
Wang, Y. (2011). Time‐delayed dynamic output feedback H∞ controller IEEE/ASME Transactions on, 20(2), 773–781.
design for civil structures: A decentralized approach through homotopic Zhou, A., Qu, B.‐Y., Li, H., Zhao, S.‐Z., Suganthan, P. N., & Zhang, Q. (2011).
transformation. Structural Control and Health Monitoring, 18(2), 121–139. Multiobjective evolutionary algorithms: A survey of the state of the art.
Wang, N., & Adeli, H. (2012). Algorithms for chattering reduction in system Swarm and Evolutionary Computation, 1(1), 32–49.
control. The Journal of the Franklin Institute, 349(8), 2687–2703. Zhu, W., & Jiang, Z.‐P. (2015). Event‐based leader‐following consensus of
Wang, N., & Adeli, H. (2015a). Self‐constructing wavelet neural network multi‐agent systems with input time delay. Automatic Control, IEEE
algorithm for nonlinear control of smart structures. Engineering Applica- Transactions, 60(5), 1362–1367.
tions of Artificial Intelligence, 41, 249–258.
Wang, N., & Adeli, H. (2015b). Robust vibration control of wind‐excited
highrise building structures. Journal of Civil Engineering and Manage- How to cite this article: Gutierrez Soto M, Adeli H. Recent
ment, 21(8), 967–976. advances in control algorithms for smart structures and
Wang, Y., & Inman, D. J. (2012). A survey of control strategies for simulta- machines. Expert Systems. 2017;34:e12205. https://doi.org/
neous vibration suppression and energy harvesting via piezoceramics.
Journal of Intelligent Material Systems and Structures, 23(18), 2021–2037.
10.1111/exsy.12205

Wang, C., & Lin, Y. (2015). Decentralized adaptive tracking control for a
class of interconnected nonlinear time‐varying systems. Automatica,
54, 16–24.

You might also like