PSO with sharing for multimodal function optimization

The particle swarm optimizer (PSO) has exhibited good performance on unimodal problem. But on multimodal problem it tends to suffer from premature convergence. In this paper, we propose a modification of the base PSO named fitness sharing particle swarm optimizer (FSPSO) to solve such problem. Through using fitness sharing technique, FSPSO can maintain diversity and avoid convergence to local solution successfully. Experiment indicates that FSPSO is an effective strategy on multimodal problem.