基于遗传算法的模式搜索法求解地下水管理模型  

Pattern Search Method for Solving Groundwater Management Model based Genetic Algorithm

作　　者：周念清[1] 陈剑桥[1] 江思珉[1] 

机构地区：[1]同济大学水利工程系,上海市200092

出　　处：《勘察科学技术》2011年第1期18-21,24,共5页Site Investigation Science and Technology

基　　金：国家自然科学基金资助(编号:41002078)

摘　　要：为改善遗传算法局部寻优精度较差的固有缺陷,该文引入一种新的基于遗传算法的模式搜索法求解地下水管理模型。首先通过在遗传算法中引入精英保留的进化策略,提高遗传算法的收敛性,进行全局搜索得到全局近似最优解,在此基础上采用局部寻优类算法模式搜索法进一步改善优化结果,最后将该混合优化方法用于求解一个地下水管理的基准问题。计算结果表明,基于改进遗传算法的模式搜索法,改善了单一遗传算法局部寻优精度较差的缺陷,在地下水管理模型的优化求解中具有较好的应用前景。In order to overcome the inherent disadvantage of the genetic algorithm,an improved hybrid optimization method was introduced to solve groundwater management models.The improved hybrid method was based on genetic algorithm and a local search algorithm named pattern search method.Firstly,the elitist strategy was introduced into GA to improve the global convergence,and an approximate global optimal solution was obtained;secondly,the pattern search method was utilized for further improving the approximate global optimal solution;finally,the introduced hybrid optimization method was adopted to solve a benchmark groundwater management problem.The simulation-optimization results reveal that this hybrid optimization method,to some degree,can overcome the disadvantage of the genetic algorithm,and may have a good application perspective on solving groundwater management model.

关 键 词：地下水管理模型 遗传算法 精英保留策略 模式搜索法 

分 类 号：P641.8[天文地球—地质矿产勘探]