A Novel Method for Finding Global Best Guide for Multiobjective Particle Swarm Optimization

The Elitism, which is the mechanism to incorporate external useful solutions in MOEA, is popular technology. One of the focus of research about elitism is how to maintain and select the global guide in order to keep the results of algorithm convergence and diversity. In this paper, a novel method to maintain elitism archive and select global guide is proposed, which divide the non-dominated solutions in the elitism archive to two kind: convergence solution and diversity solution and provides the particle angle division to manage it. The MOPSO algorithm based on the new method is compared with other multi-objective evolutionary algorithm on three complicated benchmark multi-objective function optimization problems. It is shown from the results that the Pareto front obtained with the MOPSO has good convergence and diversity.