A Hybrid Multi-swarm Co-evolutional Particle Swarm Optimizer

Abstract

In this paper, a novel cooperative particle swarm optimization (CPSO) algorithm which embodies two particle swarms is proposed to alleviate the premature convergence problem of PSO algorithm. The underlying idea of this approach is to utilize random mutation, multi-swarms, and the hybrid of many heuristic optimization methods. Firstly, an improved PSO is proposed, which adopts a new learning scheme and a random mutation operator. Then, the two swarms execute IPSO independently to maintain the diversity of populations, after a certain iteration intervals, extremal optimization (EO) and simulated annealing (SA) are introduced to the two swarms separately. By cooperative exchanging information and the hybrid of global exploration ability of PSO, local exploitation of EO and the statistical promise to deliver a globally optimal solution of SA, the performance of the traditional PSO with single swarm is improved. Simulations on a suite of benchmark functions clear demonstrate the superior performance of the proposed algorithm in terms of solution quality, convergence time.