摘要
粒子群算法的惯性因子是算法中的一个重要的参数,目前的研究结果表明,惯性因子为减函数时算法的运行效果更为良好。文中提供了四种减函数作为惯性因子可以使用的算子,它们的凹凸性各有不同。对四个算例的数值仿真结果表明,表现最好的是惯性因子先上凸后下凸的PSO,惯性因子为下凸函数的PSO综合表现优于惯性因子为上凸函数的情况。
Aimed to the efficiency changes of PSO caused by different inertia weight operators, some research and analysis had been done in this paper. The study showed that the inertia weight should decrease progressively if you want to expand the search region and assure the convergence of PSO. Four operators of inertia weight were proposed in this paper,their convexity were different with each other.The research about four examples showed that if the inertia weight operator was concave at first and then went to convex, the performance of corresponding PSO was best in all four circumstances, and the convex strategy performed better than concave strategy.
出处
《价值工程》
2015年第20期198-200,共3页
Value Engineering
关键词
粒子群算法
惯性因子
凹凸性
收敛
Particle Swarm Optimization
inertia weight
convexity
convergence
