一种求解大规模线性方程组的混合粒子群算法  

A Hybrid Particle Swarm Optimization for Solving Large-Scale Linear Equation Systems

作　　者：蒋英钰[1] 孙宁[1] 

机构地区：[1]桂林电子科技大学机电工程学院,广西桂林541004

出　　处：《计算机工程与科学》2009年第2期64-65,72,共3页Computer Engineering & Science

基　　金：国家自然科学基金资助项目(50365003)

摘　　要：本文介绍了一种大规模0/1线性方程组的特点,以及用标准粒子群算法求解时出现的不足。为此,提出了一种混合粒子群算法。该算法引入了遗传算法的变异机制,采用自适应惯性权重,动态调节粒子搜索时间,克服了标准粒子群算法求解该类线性方程组时易早熟、收敛精度低的缺点。仿真实验结果表明,采用混合粒子群算法能够有效地求解该类线性方程组。The characteristics of a type of mass 0/1 linear equation systems and the shortages of the standard particle swarm optimization algorithm are introduced in this paper. So, a hybrid particle algorithm is presented. The algorithm introduces the mutation mechanism of GA, and adopts the adaptive inertia weight to dynamically adjust the particle search time. Thus the drawbacks of easy prematurity and the low convergence precision of the standard particle swarm algorithm are overcome. The simulation experiment shows that adopting the hybrid particle swarm optimization algorithm can efficiently solve this type of linear equation systems.

关 键 词：线性方程组 混合粒子群算法 遗传算法 自适应惯性权重 变异机制 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]