A multiagent genetic particle swarm optimization

Abstract

The efforts of this paper are proposing a multi-agent genetic particle swarm optimization algorithm (MAGPSO) by introducing the multi-agent system to the particle swarm optimization(PSO) algorithm. Through the competition and cooperation operation with its neighbors, the neighborhood random crossing operation within its neighboring area, the mutation operation, and combining the evolutionary mechanism of the PSO algorithm, every individual senses local environment unceasingly, and affects the entire agent grid gradually, so that it enhances its fitness to the environment. This algorithm can maintain the diversity of the swarm effectively, and improve the precision of optimization, and simultaneously, restrain the prematurity phenomenon efficiently. The results of testing three high dimension benchmark function and comparing with some optimization results of other methods illustrate this algorithm has higher optimization performance in the field of high dimension functions optimization. © 2008 Springer Berlin Heidelberg.