Particle Swarm Optimization via Convergence-Divergence Mechanism
Jianguo Jiang · Journal of Information and Computational Science · 2015
Based on the convergence behavior of Particle Swarm Optimization (PSO), the ways that how the global best particle jump out of current state are studied. An improved PSO is proposed to solve the problems which exist in the classical PSO, such as hard to jump out of the local optima once fall into prematurity, low convergence precision and so on. The mechanism of \convergence-divergence is proposed so that the population may jump out of prematurity state through the divergence; replacement of global best position via information exchange is proposed for making full use of particles’ information and flnding better locations of particles; an improved self-adaptive exploratory mobile method is proposed for improving the convergence precision. Some classical test functions are used in experiment. The simulation results show that the new algorithm improves the convergence precision efiectively.