摘要
针对粒子群算法进行多极点函数优化时存在的局部极小点和搜寻效率低的问题,引入了小生境的思想到粒子群算法中,以粒子的最好位置为中心,粒子的最好的个体解对应的适应值为半径建立圆形小生境.在每个小生境中对粒子的速度位置进行更新,从而改变小生境的中心和半径,直到满足迭代次数,从而保持了微粒群的多样性,通过一个经典函数进行仿真表明,这种把粒子群和小生境结合起来的算法,能快速有效地找到多峰函数的全局最优点.
Particle swarm optimization （PSO） algorithm is easy to be trapped into local minima and has low searching efficiency in optimizing multimodal function. The niche algorithm is proposed into PSO in this paper. It takes the best position of particles as niche＇s center,and takes fitness value of the best particle as radius,then update the velocity and position of each particle of every niche,so a new center and radius of niches are created, until stratify the request of iterate times, so the multiplicity of particles is remained, the experiments indicated that the niche PSO algorithm can seek the global optimal value quickly and high efficiently.
出处
《机械与电子》
2007年第1期58-60,共3页
Machinery & Electronics
基金
国家科技攻关计划子课题(2004BA204B08-03)
关键词
粒子群
小生境
多峰函数
全局优化
particle swarm
niche
multimodal function
global optimize
