一种求解高维约束优化问题的γ-PSO算法  

γ-PSO algorithm for solving high-dimensional constrained optimization problems

作　　者：张慧斌[1] 王鸿斌[1] 邸东泉[1] 

机构地区：[1]忻州师范学院计算机科学与技术系,山西忻州034000

出　　处：《计算机工程与应用》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)

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.

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 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]