-
Genetic algorithms are search and optimisation methods based on the mechanics of
natural evolution. They provide promising results in non-linear and complex search
problems. They have been proven to be parallel and global in nature. However, genetic
algorithms run slowly on sequential machines. The reason for that is that the
sequential computation does not reflect the true spatial structure of the algorithm.
This research has developed a new parallel model of genetic computation which improves on existing parallel models and is suitable for real-time applications in system identification and intelligent control. Using the Markov chain formalism the model has been shown to be global convergent. The salient features of this new model are: - independence of the optimisation problem; - ablility to cope with a black box problem; - suitability for real-time applications in which the exact model of the system is unknown; and implementability within the current technological framework. The details of the model are provided in a separate publication.
Gradient like information has been integrated into the genetic search in order to improve the performance and efficiency of the algorithm. A novel directional search method has been developed and shown to result in significantly improved performance of genetic optimisation.
Unlike neural networks and fuzzy systems, genetic algorithms do not provide
any general logic for system modelling. Therefore system identification is achieved by
means of fuzzy network for general logic and a genetic algorithm for parameter
estimation giving as a result an evolving fuzzy network. This novel method has been
applied to modelling of chaotic time series and it has been used to control a highly
non-linear system, i.e. inverse pendulum. It is expected that with the advance of
re-configurable electronics, evolutionary chips will be realised in the near future
and they will fully exploit the potential of the genetic algorithms based
control systems.
Simulation Modelling - home page