Multimodal function optimizing by a new hybrid nonlinear simplex search and particle swarm algorithm

Abstract

A new hybrid Particle Swarm Optimization (PSO) algorithm is proposed in this paper based on the Nonlinear Simplex Search (NSS) method for multimodal function optimizing tasks. At late stage of PSO process, when the most promising regions of solutions are fixed, the algorithm isolates particles that fly very close to the extrema and applies the NSS method to them to enhance local exploitation searching. Explicit experimental results on famous benchmark functions indicate that this approach is reliable and efficient, especially on multimodal function optimizations. It yields better solution qualities and success rates compared to other three published methods. © Springer-Verlag Berlin Heidelberg 2005.