粒子群优化中最大速度选择

Maximun Velocity Selection in Particle Swarm Optimization

文中研究了粒子群优化中最大速度与优化性能的关系,选择了四个具有非对称初始化设置的基准函数作为测试函数。实验结果表明把最大速度设定在动态范围内的一个较小区间,粒子群优化性能最佳。为了便于描述,引进了最佳速度因子u,分析并测试了不同参数下u对优化性能的影响,为最大速度的选择提供了有效的指导。仿真结果表明,最佳速度因子u是优化过程中的一个特征参数,它随着粒子维数的升高而有所增加,对粒子的个数和初始化范围并不敏感。

The performance of the particle swarm optimization under different maximum velocity is investigated in this paper. Four different benchmark functions with asymmetric initial range settings are selected as testing functions. The experimental results show that the optimal performance of particle swarm optimization is obtained by setting a small maximum velocity allowed. For being easy to describe, maximun velocity factor （u） is introduced. The effect of maximum velocity factor on particle swarm optimization is tested and analyzed, which provides guidelines for selecting maximum velocity. The result of emulation indicates that maximun velocity factor （u） is a characteristic parameter of the optimization process, it increases with the increasing of partide dimension, isn＇ t sensitive to the amount of particle and its initialization range .