摘要
PSO算法是一种随机搜索的群体智能算法,在求解高维约束优化问题,尤其是在约束条件较多时,PSO算法易陷入局部极值且收敛速度慢。针对上述问题,对PSO算法进行了改进,提出了γ-PSO算法,把PSO算法的随机数由(0,1)扩展到(-1,1),这样加大了粒子飞行速度和飞行方向的多样性,从而使PSO算法具有摆脱局部极值的能力。对γ-PSO算法进行了求解高维约束优化问题的实验,实验结果表明γ-PSO算法能收敛到全局最优值,收敛性能明显优于其他改进的PSO算法和其他优化算法。
PSO algorithm is one of random searching swarm intelligence algorithm for solving multi-dimensional constrained optimization problem. But when the constraints become more, PSO algorithm is easy to fall into local minimum and slow convergence. In response to these problems, γ-PSO algorithm is proposed, an improved PSO algorithm, which extends random numbers from（0, 1）to（1, 1）. In this way, the PSO algorithm can avoid local minimum by increasing flying speed and diversity of flying direction of particle. Finally, the results of experiments using γ-PSO algorithm for solving high-dimensional constrained optimization problems show that the γ-PSO algorithm can converge to the global optimum, and its convergence is superior to other improved PSO algorithms and other optimization algorithms.
出处
《计算机工程与应用》
CSCD
2012年第7期43-47,83,共6页
Computer Engineering and Applications
基金
山西省自然科学基金(No.2009011018-4)
关键词
PSO算法
约束优化问题
适应度函数
全局极值
局部极值
Particle Swarm Optimization（PSO）algorithm
constrained optimization problem
fitness function
global optimum
local minimum
