Computational Intelligence: An Introduction - Computational Intelligence in Time Series Forecasting - page 9

Evolutionary algorithms have also been successfully used in combination with
fuzzy logic in improving heuristic rules and in manipulating optimally the genetic
parameters, particularly the crossover operator (Herrera and Lozano, 1994).
Neural networks, in combination with evolutionary algorithms, have profited in
optimal evolution of network topology and in finding the optimal values of
network weights directly, without network training (Maniezo, 1994). Finally,
evolutionary algorithms have also profited through combinations with the
traditional computing methods. For instance, in order to improve the efficiency and
the accuracy of evolutionary computing algorithms in locating the global
extremum, Renders and Bersini (1994) combined these algorithms with the
conventional search methods, such as the hill climbing method. Renders and Flasse
(1976) even simply integrated such a method in the crossover operator.
1.7 Application Areas
Computational intelligence and soft computing have proven to be very efficient
and valuable tools for solving numerous problems in science and engineering that
could not be solved using their individual constituents, i.e. neuro, fuzzy, or
evolutionary computing alone. Although their constituents are themselves capable
of solving problems that are difficult or even impossible to solve by traditional
computation methods, the synergetic effect of aggregation of two or more
constituents enlarges the number and the complexity of solvable problems. This
holds not only for the so-called academic problems, but also for real-life problems,
including the problems of industrial engineering. Moreover, application of soft
computing and computational intelligence has provided the appropriate means for
merging the vagueness ( e.g. perceptions of human beings) and real-life uncertainty
with a relatively simplified computational program. This has made them capable of
participating in a variety of real-life applications in engineering and industry. For
instance, the application of soft computing in engineering covers most areas of data
handling, like:
x intelligent signal processing, which includes time series analysis and
forecasting
x data mining
x multisensor data fusion, including intelligent pattern recognition and
interpretation, performance monitoring and fault diagnosis
x systems engineering, to which belong system identification, system
modelling, advanced systems control
x planning and design processes, like optimal path planning and engineering
design
Intelligent signal processing solves the problems of adaptive signal sampling,
analysis of sampled data, signal features extraction, etc. Of outstanding interest for
engineering, commerce, and management here is the forecasting of time series data
(Kim and Kim, 1997).