基于单纯形的小生境混合遗传算法  

Niche Hybrid Genetic Algorithm Based on Simplex

作　　者：肖宏峰[1] 谭冠政[2] 

机构地区：[1]中南大学信息科学与工程学院机器人研究所,湖南长沙410083 [2]湖南师范大学计算机教学部,湖南长沙410081

出　　处：《小型微型计算机系统》2008年第9期1719-1725,共7页Journal of Chinese Computer Systems

基　　金：国家自然科学基金项目(5027150)资助;高等学校博士点基金项目(20040533035)资助

摘　　要：总结单纯形搜索算法的核心思想.然后提出单纯形交叉方向算子和最优小生境、次差小生境与最差小生境3个概念.在最优小生境中采用单纯形搜索算法得到局部极值,在最优小生境与次差小生境之间用单纯形交叉方向算子产生优秀个体,而在最差小生境中采用受限单纯形搜索产生优秀个体,从而构成基于单纯形的小生境混合遗传算法SimplexNich-HGA.最后用Sim-plexNiche-HGA、单纯形混合遗传算法Simplex-HGA+以及基本遗传算法SGA求函数Rosenbrock的极值,并进一步用Sim-plexNiche-HGA和Simplex-HGA+求多峰值函数Shubert的极值,验证算法的正确性和求多峰值函数的极值的效率.Conclude the kernel idea of simplex search algorithm. Propose simplex crossover direction operator and three concepts of the best niche, worse niche and the worst niche. In the best niche utilize simplex search algorithm to obtain local extreme value, between the best niche and worse niche, put to using simplex crossover direction operator to produce some better chromosomes, and while in the worst niche,wield limited simplex search to produce better chromosomes. Build a niche hybrid genetic algorithm based on simplex, which is called SimplexNiche-HGA. At last ,wield SimplexNiche-HGA, simplex hybrid genetic algorithm Simplex-HGA^＋ and simple genetic algorithm SGA to obtain the extreme value of test function Rosenbrock,and further to obtain the extreme values of test function Shurtert by algorithms of SimplexNiehe-HGA and Simplex-HGA^＋ ,all simulation results show that SimplexNiche-HGA is correct and more efficient than Simplex-HGA^＋.

关 键 词：单纯形搜索算法 单纯形交叉方向算子 混合遗传算法 小生境遗传算法 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]