A hybrid multi-swarm PSO algorithm based on shuffled frog leaping algorithm

Abstract

As an effective swarm intelligence algorithm, multi-swarm particle swarm optimization (PSO) has better search ability than single-swarm PSO. In order to enhance the ability of group communication as well as improve the ability of local search, this paper proposes a hybrid multi-swarm PSO algorithm. Three strategies have been proposed, which are multi-swarm strategy, update strategy and cooperation strategy. A new way of grouping the particle swarms is put forward by calculating the fitness value of particles. In each group, the particles updates according to the formula which is morphed from the shuffled frog leaping algorithm. Moreover, a new information communication strategy is proposed. The cooperation of these three strategies maintains the diversity of algorithm and improves the ability of searching the optimal solution. Finally, the experimental results on the benchmark functions verify the effectiveness of the proposed PSO.