Enhancing Particle Swarm Algorithm for Multimodal Optimization Problems

Abstract

Particle swarm optimization (PSO) is an intelligent algorithm inspired by swarm intelligence. It has been shown that PSO is a good optimizer on various optimization problems. Due to the inherent randomness of PSO, it easily falls into local minima when dealing with multimodal optimization problems. In order to enhance the performance of PSO on multimodal problems, this paper proposes a novel PSO algorithm by employing adaptive parameter control and example-based learning. Conducted experiments on nine well-known multimodal problems show that our approach outperforms the standard PSO, unified PSO (UPSO), fully informed PSO (FIPS), fitness-distance-ratio based PSO (FDR-PSO), cooperative PSO (CPSO-H) and comprehensive learning PSO (CLPSO) in terms of the solution accuracy.