Hybrid Particle Swarm Optimiser using multi-neighborhood topologies

Abstract

Hybrid Particle Swarm Optimization (PSO) algorithm that combines the idea of global best model with the idea of local best model is presented in this paper. The hybrid PSO mixes the use of the traditional velocity and position update rules of star, ring and Von Neumann topologies all together. The objective of building PSO on multi-models is that, to find a better solution without trapping in local minimums models, and to achieve faster convergence rate. This paper describes how the hybrid model will get the benefit of the strength of gbest and lbest models. It investigates when it would be better for the particle to update its velocity using star or ring or Von Neumann topologies. The performance of proposed method is compared to other standard models of PSO using variant set of benchmark functions to investigate the improvement.