基于单纯形的小生境混合遗传算法

英文篇名：Niche Hybrid Genetic Algorithm Based on Simplex
中文刊名：小型微型计算机系统
英文刊名：Journal of Chinese Computer Systems
作者：肖宏峰 ; 谭冠政
英文作者：XIAO Hong-feng1 ; 2 ; TAN Guan-zheng1 1(Robotic Institute ; School of Information Science & Engineering ; Central South University ; Changsha 410083 ; China) 2(Computer Education Department of Hunan Normal University ; Changsha 410081 ; China)
中文关键词：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
英文关键词：simplex search algorithm ; simplex crossover direction operator ; hybrid genetic algorithm ; niche genetic algorithm
出版日期：2008-09-15
机构：中南大学信息科学与工程学院机器人研究所;湖南师范大学计算机教学部;
年：2008
期：09
出版单位：小型微型计算机系统

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 SimplexNiche-HGA and Simplex-HGA+,all simulation results show that SimplexNiche-HGA is correct and more efficient than Simplex-HGA+.

引文

[1]Oliver L M.A study of permtutaion crossover operators on thetraveling salesman problem[C].Proc.of 2th Int.Conf.on Ge-netic Algorithms,1987,224-230.
    [2]Tan Yong-ji,Zeng Yi,Cao Wei.Solution to NLP by using hy-brid genetic algorithms[J].Journal of Fudan University(Natu-ral Science),2000,39(5):465-470.
    [3]Mahfoud S W,Goldgerg D E.A genetic algorithm for paralletsimulated annealing[J].Parallet Problem Solving from Nature2,North Holland,1992,26(3):301-310.
    [4]Joln Yen,James C,Liao Bojwlee.A hybrid approach to modelmetabolic system using a genetic algorithm and simple method[J].IEEE Transactions on System,Man and Cybernetic-part B:Cybernetic,1998,28,(2):.
    [5]He Da-kuo,Li Yan-qiang.Hybrid genetic algorithm based onoperator of pattern search[J].Information and Control,2001,28(3):276-278.
    [6]Han Wei,Liao Zhen-peng.A global optimization algorithm:ge-netic-simplex[J].Earthquake Engineering and Engineering Vi-bration,2001,21(2):6-12.
    [7]Shen Jie,Wang Xin.Application of improved simplex genetic al-gorithm hybrid in automatic optical fiber algnment[J].Opto-Electronic Engineering,2006,33(10):67-71.
    [8]Yan Wei,Zhao Da.A new optimization algorithm based on theprinciple of evolution[J].Journal of Electronics,1998,15(2):248-253.
    [9]Lee S,Olson D.A gradient algorithms for chance constrainednonlinear programming[J].European Journal of Operational Re-search,1977,11(1):343-369.
    [10]Zhou Ming,Sun Shu-dong.Principle and application of genteticalgorithm[M].Beijing:Publish house of National Defense Tech-nical,1999.
    [2]谭永基,曾毅,曹维.用混合遗传算法求解约束NLP问题[J].复旦大学学报(自然科学版),2000,39(5):465-470.
    [5]何大阔,李延强.基于单纯形的混合遗传算法[J].信息与控制,2001,30(3):276-278.
    [6]韩炜,廖振鹏.一种全局优化算法:遗传算法-单纯形法[J].地震工程与振动工程,2001,21(2):6-12.
    [7]沈洁,王昕.改进遗传单纯形混合遗传算法在光纤对接中的应用[J].光电工程,2006,33(10):67-71.
    [10]周明,孙树栋.遗传算法原理及应用[M].北京:国防工业出版社,1999.