Geology Reference
In-Depth Information
and sensors in structures. Li et al. (2000; 2004)
developed a multi-level GA to optimize both ac-
tuator locations and state feedback control gains
for structural control system design. GA is also
applied to a forty-story high-rise building to find
the optimal locations of the pre-defined number
of actuators (Abdullah et al. 2001). Cheng et al.
(2002) applied a sequential iterative procedure
for optimal placement of dampers and actuators
to a seismically excited three-story building. A
step-by-step procedure for optimal placement of
piezoelectric friction dampers in a seismically
excited building is proposed by Chen and Chen
(2002; 2004). Liu et al. (2003) adopted GA for
optimal actuator distribution within a seismically
excited sixteen-story tall building. Wongprasert
and Symans (2004) proposed an optimal loca-
tion of passive control devices within the ASCE
nonlinear benchmark building. Yang et al. (2005)
applied SA to an optimization problem of finding
best locations of active bars in smart structures.
Amini and Tavassoli (2005) applied artificial
neural networks to the optimal actuator place-
ment problems for seismic response control of a
twelve-story building structure. Tan et al. (2005)
applied GA to optimization problems of finding
optimal actuator locations and control gains for
hazard mitigation of a forty-story shear building
and a nine-story irregular structure. Moita et al.
(2006) applied an SA to laminated reinforced
composite structures to maximize the effective-
ness of piezoelectric actuators. Rao and Sivasub-
ramanian (2008) proposed a novel multiple start
guided neighborhood search (MSGNS) algorithm
by integration of the best features of SA and Tabu
search algorithms for optimal placement of ac-
tuators within seismically excited tall buildings.
However, there is minimal study of GA-based
multi-objective optimal formulations for mini-
mum distributions of both actuators and sensors
as well as minimum structural responses of large-
scale infrastructures under seismic excitations.
Several previous studies used a simple GA to
solve this optimal distribution of the actuators
and sensors. However, it might not be easy to
solve highly complex optimization problems, e.g.,
multi-objective formulations of large-scale com-
plex structure-control systems. Even though it can
handle the complex problem, it may require high
computational cost (Raich and Ghaboussi 1998).
With this in mind, this topic chapter introduces
three novel multi-objective genetic algorithms
(MOGA) with the capacity of robust and efficient
problem solving for optimal placement of control
devices and sensors in large civil structures such
that the performance on the interstory drifts of
structures is also satisfied: 1) the proposed first
MOGA is developed through the integration of
an implicit redundant representation genetic al-
gorithm (IRR GA) (Raich 1999) and a strength
Pareto evolutionary algorithm 2 (SPEA2), namely,
SP2-IRR GA (Cha et al. 2011a) ; 2) the second
one is an integrated model of a non-dominated
sorting genetic algorithm 2 (NSGA-II) (Deb et
al. 2000) and IRR GA, namely, NS2-IRR GA
(Cha et al. 2011b); 3) gene manipulation genetic
algorithm (GMGA) (Cha et al. 2011a) by applying
engineering judgment concept as a genetic opera-
tor. To investigate the effectiveness of the newly
proposed algorithms, full-scale twenty-story
buildings are investigated. To implement active
structural control systems into the large frame
structures, the linear quadratic Gaussian (LQG)
algorithm, hydraulic actuators, and accelerometers
are used. In this topic chapter, multi-objective
optimization problems are formulated using two
tradeoff objective functions of the number of
actuators and sensors, and the interstory drifts of
the frame structures.
MULTI-OBJECTIVE GENETIC
ALGORITHMS
The basic idea of a simple GA (SGA), which is
developed by Goldberg (1989), coming from
natural selection in Darwin's theory is composed
of four steps: 1) evaluation of each individual's
Search WWH ::




Custom Search