改进惯性权重的粒子群算法及其在边坡稳定性分析中的应用  

A modified particle swarm optimization(PSO) algorithm with dynamic inertia weight and its application to slope stability analysis

作　　者：蔡德所[1,2] 吴彰敦[1] 陶俊波[1] 邱飞 

机构地区：[1]广西大学土木建筑工程学院,广西南宁530004 [2]三峡大学土木水电学院,湖北宜昌443002 [3]玉林市水利电力科学研究所,广西玉林537000

出　　处：《水利水电科技进展》2009年第1期20-22,51,共4页Advances in Science and Technology of Water Resources

摘　　要：改进PSO算法的惯性权重。不仅考虑了惯性权重随代数的纵向线性变化,还根据当前和迄今粒子的适应度重排序横向线性变化来改进惯性权重。横向线性变化上限不变,下限逐渐减小,使得线性变化数值范围随代数逐渐增大。惯性权重随着代数逐渐取负,并且适应度差的粒子取负的几率更大。这样得到基于粒子适应度排序改进惯性权重的粒子群算法(ASMIWPSO算法)。通过2个仿真算例对比ASMIWPSO算法和PSO算法的寻优结果,所获得的全局最优值前者多于后者。采用边坡工程实例进行ASMIWPSO算法、PSO算法和理正岩土计算软件结果比较,说明ASMIWPSO算法具有更好的优化结果。In this paper, the inertia weight of the PSO algorithm is modified, not only in terms of its longitudinal linear change with iteration times, but also in terms of its lateral linear change, because of an adaptation sequence of the current particle and the optimal particle. The numerical range of inertia weight increases with the iteration times, with the upper limit of the lateral linear change keeping constant while the lower limit decreasing gradually. Hence, this modified PSO algorithm with dynamic inertia weight, which is known as the ASMIWPSO algorithm, is obtained. Through two simulation instances, the optimization effect of the ASMIWPSO algorithm is cempared with that of the PSO algorithm. Then, the ASMIWPSO algorithm, the PSO algorithm and LI Zheng software are applied to a practical slope engineering problem, whose results indicate that the ASMIWPSO algorithm has a better optimization effect.

关 键 词：粒子群算法 惯性权重 适应度排序 最小安全系数 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构] TU31[自动化与计算机技术—计算机科学与技术]