Abstract: Dual population genetic algorithm with chaotic local search strategy is proposed to solve bad local search ability and early convergence which are the two defects of genetic algorithm. In proposed algorithm, one population is used as exploration population, the other is exploitation population. The two populations are evolved by different crossover probability and mutation probability. At the end of each generation, chaotic local search is applied to the optimal solution of each population, and the solution will be the new optimal solution if a solution found by chaotic local search is better than the optimal solution. Chaotic local search is not stopped until the predefined search time is elapsed. An immigration operation is down between the two populations each ten generation. Experimental results on six benchmark functions show that proposed algorithm had the better ability of finding optimal solution.

Key words: chaoic search, local search, premature convergence, dual population genetic algorithm, function optimization

摘要： 针对遗传算法局部搜索能力差和早熟收敛的问题，提出一种基于混沌局部搜索的双种群遗传算法。将2个种群分别作为探测种群和开发种群，按不同交叉概率和变异概率进化。种群每进化一代即对其最优解做混沌局部搜索，若搜索到更优解，则取代原最优解，直至搜索到预设的混沌次数，同时2个种群之间每进化10代进行一次移民操作。在6个Benchmark函数上的实验结果表明，该算法具有较好的寻优能力。

关键词: 混沌搜索, 局部搜索, 早熟收敛, 双种群遗传算法, 函数优化

CLC Number: 

• TP18