基于改进模式提取变异算子的遗传算法  

An Improved Algorithm of Pattern Extraction Based Mutation Approach to Genetic Algorithm

作　　者：戴晓明[1] 陈治纲[1] 冯瑞[1] 茅雪飞[1] 邵惠鹤[1] 

机构地区：[1]上海交通大学自动化研究所,上海200030

出　　处：《上海交通大学学报》2002年第8期1158-1160,共3页Journal of Shanghai Jiaotong University

基　　金：国家重点基础研究发展规划 (973 )项目 (G19980 3 0 4)

摘　　要：在经典遗传算法的基础上 ,提出了一种基于改进模式提取 ( Algorithm of pattern extrac-tion,Alopex)的变异算子 ,种群个体的连续进化方向作为当前代个体的变异方向 ,并利用自适应来调整变异步长 ,通过控制参数来控制变异方向的概率从而跳过局部最优值 .对几种典型函数的测试结果表明 ,基于该变异算子的遗传算法能较好地避免收敛到局部最优 。This paper presented an Algorithm of pattern extraction (Alopex) approach to the mutation mechanism for Genetic Algorithm (GA). The direction of the mutation of individual is guided by two consecutive evolve directions of the individual. The self adaptive step is adjusted by the two consecutive improved directions. The probability of overcoming the local minimum is determined by certain control parameters. Several benchmark problems were tested for the new approach's efficiency and convergence property. The plaguing problem of premature convergence problem can be greatly levied by this approach. It shows great advantage to the canonical genetic algorithm.

关 键 词：变异算子 遗传算法 模式提取 收敛性 导向变异能力 局部最优 

分 类 号：O242.23[理学—计算数学]