A Hybrid Particle Swarm Optimization Algorithm for Function Optimization

摘要

In this paper, a new variation of Particle Swarm Optimization (PSO) based on hybridization with Reduced Variable Neighborhood Search (RVNS) is proposed. In our method, general flow of PSO is preserved. However, to rectify premature convergence problem of PSO and to improve its exploration capability, the best particle in the swarm is randomly re-initiated. To enhance exploitation mechanism, RVNS is employed as a local search method for these particles. Experimental results on standard benchmark problems show sign of considerable improvement over the standard PSO algorithm.