A cultural particle swarm optimization algorithm

摘要

A new culture-based Particle Swarm Optimization (PSO) with mutation, CPSOM, is proposed in this paper to improve the overall optimization performance of the original PSO and combat with the well-known premature problem. In the CPSOM, the Evolutionary Programming (EP) mutation operator is applied to a proportion of the particles in the population space based on the influence function. The mutation operation is directed by the knowledge stored in the belief space, and the mutation proportion can vary linearly with the growth of the swarm generations. Our CPSOM is investigated using ten high-dimension and multi-peak functions. Numerical simulation results demonstrate that it can indeed outperform both the original PSO and EP.