Dynamic Particle Swarm Optimization Based on Neighborhood Rough Set Model

To obtain the prior space information on the study problems and prevent the blind search, a novel strategy on the particle swarm optimization (PSO) is proposed. Based on the neighborhood rough set model, the prior information is achieved to guide the evolutionary state of the PSO constantly. By reserving the much relevant area of the global best point, the search space was dynamically reduced. Comparison studies with another improved PSO were performed. The experimental results for most test functions demonstrated good performance of the proposed method in both the optimization speed and computational accuracy. The results are firmly verified the effectiveness of the method to obtain the prior space information and improve the performance of the PSO.