An Improved Particle Swarm Optimization

Abstract

Particle swarm optimization (PSO) has shown good search ability on many optimization problems. However, PSO easily suffers from local optima on some complex problems, such as multimodal function problems. This paper presents an improved PSO, namely IPSO, which employs an adaptive chaotic mutation operator. The adaptive mutation adjusts the step size of mutation in terms of the distance between the current particle and the global best particle. Experimental results on six wellknow benchmark functions show that IPSO performs better than the standard PSO, genetic algorithm and PSO with chaos (CPSO) on most test problems.