Review and Improvement of Several Optimal Intelligent Pitch Controllers and Estimator of WECS via Artificial Intelligent Approaches - Complex System Modelling and Control Through Intelligent Soft Computations - page 497

Multivariable control describes an approach that afford pitch angle (Laks et al.
2009 ). On the other hand, informed works exploit such adaptive controllers for
ef

cient power conversion, related to maximum power point (MPP) tracking
(Boukhezzar et al. 2007 ; Xing et al. 2009 ), with less regard to the WECS s

structural integrity. In Bianchi et al. ( 2006 ), a feature-extraction algorithm, a fre-
quency analyzer, was developed, and the features are formulated as the inputs of an
arti

'

cial previous term neural network next term using back propagation. An

arti cial-previous term neural-network next term-based controller has been pre-

sented in Kuo ( 1995 ), to realize fast valuing in a power-generation plant.
Progression of hybrid architectures (Yingduo et al. 1997 ), knowledge acquisition
for symbolic AI systems and improved adequacy for data mining applications
(Horiuchi and Kawahito 2001 ; Bianchi et al. 2006 , 2008 ), these show some purposes
of rule extraction approaches are data exploration. Genetic algorithms (GAs) have
been used in various problems, such as nonlinear optimization, combinatorial
optimization and machine learning (Muhando 2008 ; Gen and Cheng 1997 ; Goldberg
1989 ; Andrew and Haiyang 2010 ). Also genetic algorithms are applied for selecting
fuzzy if-then rules, modi cation of nonlinear scaling functions, and for determining

hierarchical structures of fuzzy rule-based systems. A cascade GA (Prakash et al.
2011 ), a micro-GA (Heider and Drabe 1997 ) is uncommon genetic algorithm that
was used for designing fuzzy rule-based systems.
Since the wind speed exceeds its nominal value, power is regulated to the
turbine

s rated output by shifting the control objective from maximizing power
catch (Glorennec 1997 ). One of the powerful universal predictors that show very
good performance solving complex problems is Neural Network (NN). Several
purposes of these rule extraction methods are data exploration, progression of
hybrid architectures (Boukhezzar et al. 2007 ), knowledge acquisition for symbolic
AI systems and improved adequacy for data mining applications (Wermter and Sun
2000 ; Mitra et al. 2002 ; Mitra 1994 ). Three popular techniques that extract rules
from a trained NN are Neurorule, Trepan and Nefclass (Witten and Frank 1999 ;
Baesens et al. 2003 ).
Rules from each unit in a NN have been extracted by the decompositional
algorithms. The so called extracted rules are then aggregated to form the

'

final fuzzy
forecasting system. In Craven and Shavlik ( 1996 ) an online training fuzzy neural
network controls the induction generator via a high performance speed observer.
Hong et al. (Nauck 2000 ) propose a new method of wind power and speed fore-
casting using a multi-layer feed-forward neural network. They develop a forecasting
system for time-scales that might vary from a few minutes to an hour. In Lin and
Hong ( 2010 ), Lin et al. ( 2010a , b ), a Wilcoxon radial basis function network with
hill-climb searching maximum power point tracking strategy is suggested for a
permanent magnet synchronous generator with a variable-speed wind turbine. An
approach for optimization of power factor and the power produced by wind turbines
was presented in Hong et al. ( 2010 ). Data-mining algorithms capture the relation-
ships among the power output, power factor, and controllable and non-controllable
variables of a 1.5 MW wind turbine. In Lin and Hong ( 2010 ), Lin et al. ( 2010a , b ),
the modeling and the control of a WECS associated to a super capacitor module as