Modified particle swarm optimization based on optimum-selecting by probability and explosive searching
Xuexing Ming, Jing Fang Qian, Jianguo Wang, Zhenzhong Lv · 2008
To overcome the drawback of premature convergence of standard particle swarm optimization (PSO) especially when solving high-dimension functions, this paper provided a modified particle swarm optimization(MPSO) based on optimum-selecting by probability and explosive searching strategy. In each iteration, every particle selects individual gbest with optimum-selecting by probability and takes explosive searching algorithm during the path towards its gbest to search better particles in the suprasphere around itself for replacement. Three benchmark functions were selected as the test functions for computer simulation experiments. The final test results show that the MPSO can not only greatly speed up the convergence but also significantly solve the premature convergence of PSO.