基于细分变异算子策略的遗传算法

摘要

An improved genetic algorithm based on subdividing mutation (SMSGA) is proposed to overcome poor seeking optimization capabilities and premature phenomenon of the basic genetic algorithm. SMSGA divides mutation operator according to involution process into major step mutation operator and tuning best individual operator. Major step mutation operator prevents the occurrence of genetic premature phenomenon,while tuning best individual operator strengthens the ability of local search. Meanwhile,to speed up the convergence rate,SMSGA has also optimized the process of genetic algorithm by introducing the operation of routing choice. We compare SMSGA, ERGA (genetic algorithm based on elite reserves) and DMGA (genetic algorithm based on dual mutation) by selecting three typical test functions in the MATLAB platform,and experimental results show that SMSGA can effectively avoid poor seeking optimization capabilities and premature phenomenon.%针对基本遗传算法局部搜索能力不强以及早熟的问题,提出基于细分变异算子的遗传算法(Genetic Algorithm Based on Subdividing Mutation,SMSGA).SMSGA将变异算子依据进化历程分成大步前进算子和最优调教算子.大步前进算子防止遗传早熟现象的发生；最优调教算子加强局部搜索的能力.同时,为加快算法收敛速度,对遗传操作实施策略进行优化,引入了路由选择操作.选用3个典型的测试函数在MATLAB平台中对该算法与基本遗传算法以及采用双变异率的改进遗传算法进行比较分析,结果表明,SMSGA可以有效的避免遗传算法中存在的局部搜索能力差和早熟现象的出现.