遗传算法平均截止代数和成功率与种群规模之间的关系  

Relationship between Average Truncated Generation, Success Rate and Population Scale in Genetic Algorithm

作　　者：李金屏[1] 何苗[1] 杨波[1] 

机构地区：[1]济南大学信息与控制工程系,济南250022

出　　处：《系统仿真学报》2001年第z1期206-210,共5页Journal of System Simulation

基　　金：山东省自然科学基金青年基金(Q98G02151); 国家自然科学基金(69902005)资助项目

摘　　要：摘 要:遗传算法的搜索效率在很大程度上取决于运行参数,如种群规模M、编码串长度l、交叉概率Pc、变异概率Pm、终止代数T、适应度函数f(M)等的设置。利用简单遗传算法和统计分析方法,通过对四个典型数学结构进行全局最优解搜索,得到了遗传算法平均进化截止代数T和成功率R与种群规模M之间的关系,给出了相应的拟合公式。讨论了平均进化截止代数和成功率与种群规模和极值个数之间的关系。这种研究对于合理确定种群规模和进化终止代数,正确估计搜索到的满意解的最优程度等具有重要意义。The efficiency of genetic algorithm is dependent upon such parameters as population scale M, code length l, crossover probability Pc, mutation probability Pm, termination generation T and fitness function f (M) to a great extent. By means of simple genetic algorithm and statistical analysis, global optimums of 4 typical mathematical problems are searched, relationship between average truncated generation T, success rate R and population scale M in genetic algorithm are obtained, the corresponding fitting formula are presented. Finally, the relationship between average truncated generation, success rate and the number of optimums is discussed. The research is of great importance for reasonably determining population scale and termination generation, also for assessing the satisfactory solution.

关 键 词：遗传算法 种群规模 平均截止代数 成功率 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]