具有混沌局部搜索策略的双种群遗传算法

Dual population genetic algorithm with chaotic local search strategy

为提高遗传算法的局部和全局搜索能力,提出了一种具有混沌局部搜索策略的双种群遗传算法(CLS-DPGA)。CLSDPGA中,一个作为探测种群,另一个作为开发种群。两个种群按照不同交叉概率和变异概率进行进化,每个种群每进化一代后就对其最优解进行混沌局部搜索。若搜索到更优的解,则取代原最优解直至搜索到预设的混沌次数,同时两个种群之间每10代进行一次移民操作。六个Benchmark函数的实验结果证明,CLS-DPGA比另一种自适应局部搜索策略的遗传算法(a-hGA2)具有更好的寻优能力。

This paper proposed dual population genetic algorithm with chaotic local search strategy（CLSDPGA） to improve local and global search ability of genetic algorithm.In CLSDPGA,one population was used as exploration population,the other was exploitation population.The two population was evolved by different crossover probability and mutation probability.At the end of each generation,applied chaotic local search to the optimal solution of each population,and the solution would be the new optimal solution if a solution found by chaotic local search was better than the optimal solution.Chaotic local search was not stopped until the predefined search time was elapsed.An immigration operation was down between the two population each ten generation.Experiment results on six benchmark functions show that CLSDPGA has the better ability of finding optimal solution than that of genetic algorithm with adaptive local search scheme（a-hGA2）.