Generalized Particle Swarm Optimizer with Tracking Multiple Local Optima for Multimodal Functions Optimization

This paper presents a new variation of particle swarm optimization (PSO) algorithm called generalized particle swarm optimizer (GPSO). It extends the basic learning strategy of traditional PSO and exerts the swarms to significantly improve the group learning performance. In this scheme, a particle of PSO in each dimension does not only follow its own local optima, but also follows other superior particles´ local optima with creditability. Based on our experimental verifications, the results suggest that GPSO delivers superior performance for multimodal functions optimization compared with the state-of-art PSO methods.