Particle Swarm Optimization Based on Dynamic Island Model

Particle Swarm Optimization (PSO) algorithm is a metaheuristic that has been used for solving optimization problems. In this method, many modifications have been carried out in order to improve the search performance. Furthermore, island models is a structured population mechanism used to preserve the diversity and thus to improve the population performance. The aim of this paper is to integrate dynamic island models with PSO algorithm to improve its convergence and its diversity properties where the new method is referred to as island PSO. The dynamic regulation of migrations aims to distribute the particles in the search space. The experimental results, using a set of benchmark functions show that the island model context is crucial to the PSO performance and the comparative study shows the efficiency of the integration of dynamic island models.