均匀设计在粒子群算法参数设定中的应用

针对依靠经验和试验来确定粒子群优化算法中的参数设定方法存在试验工作量大且难以得到最优的参数组合的问题,提出了一种改进的粒子群算法,该算法利用均匀设计的思想设定算法的最优组合参数,把参数设定问题描述为均匀试验设计中多因素多水平优化设计问题,从而能够以较少试验很快设定算法参数的取值,以使算法获得最优性能。仿真试验表明,利用均匀设计得到的参数组合可使粒子群算法获得稳健和高效的优化效果,说明了该方法的可行性和有效性,且可以推广到其他算法参数的设定。

Parameter Establishment of Particle Swarm Optimization Based on Uniform Design

To the problem that the establishment of key parameters of particle swarm optimization having great influence on algorithm effects is determined by experience and experiments,and leads to heavy work and makes the optimal combination of the parameters difficult to obtain,a modified particle swarm optimization is proposed.The uniform design theory is applied to obtain the optimal establishment parameters of the algorithm.The problem of establishment parameters is described as an optimization design problem of...