Return to author index

Volume: 5 Issue: 2

A Fuzzy Clustering based Particle Swarms for Multimodal Function Optimization

Jihane Alami, Lamia Benameur, Abdelhakim A. El Imrani

Abstract:
The Particle Swarm Optimization (PSO) has been increasingly used for solving complex and difficult optimization problems. However, both experiments and analysis show that the basic PSO algorithms cannot identify different optima, either global or local, and thus are not appropriate for multimodal optimization problems that require the location of multiple optima. In this context, several multimodal optimization techniques using PSO and niching concept have been reported in the literature. These models often require a priori information about niche radius and number of optima. To overcome this limitation, a new approach, based on PSO and a fuzzy clustering method, is proposed. This model uses a fuzzy clustering technique, in order to allow development of multiple sub-swarms, and then promotes simultaneous tracking of multiple peaks without requiring any prior knowledge (e.g. Niche radius, number of optima). The ability and the efficiency of the proposed approach to identify multiple optima are demonstrated using benchmark test functions.

Keywords:
Particle swarm optimization, Fuzzy clustering, Multimodal function optimization.

doi:10.5019/j.ijcir.2004.178

Full Text PDF
















^ TOP