关键词: 粒子群算法, 复合形算法, 自适应变异

Abstract: To deal with the problem of premature convergence，slow convergence velocity，a novel Particle Swarm Optimization（PSO） algorithm is proposed.At the beginning of the evolution，PSO can search global area and find the local range quickly，and then，complex method would locate the extremum in the local range rapidly.The self-adaptive mutation inertia weight is used in the whole evolvement to break away from the local extremum，which can effectively solve the premature convergence problem.The experiment results of two classic benchmark functions show that the algorithm can not only significantly improve the convergence velocity and precision in the evolutionary optimization，but also effectively enhance the global optimization power.

Key words: Particle Swarm Optimization（PSO）, complex method, self-adaptive mutation