Abstract: Mating of biological evolution generally occurs only between two individuals, and crossover operation of the traditional genetic algorithm is also carried out between two chromosomes. In the paper, the crossover operator is carried out on three or more chromosomes and the corresponding crossover operator is designed. Multi-Chromosomes Crossover Operator（MCCO） can significantly improve the optimization ability and convergence rate, but it is easy to produce premature convergence due to the fast convergence speed. Therefore, a niche algorithm with the sub population elimination tactics is designed in order to avoid the phenomenon of premature convergence. Compared with the experimental results of several genetic algorithms, it is proven that the MCCO genetic algorithm is better than the traditional genetic algorithm in the multi-apices optimization.

Key words: genetic algorithm, Multi-Chromosomes Crossover Operator（MCCO）, niche algorithm, real coded

摘要： 生物进化一般只在两个个体间进行交配，传统的遗传算法也仅在两个染色体之间进行交叉操作。使用在三个以上的染色体进行交叉操作，并设计了多染色体交叉的算子。多染色体交叉算法可显著提高算法局部寻优能力和收敛速度，但由于收敛速度过快容易产生早熟现象。因此，设计了一种带子种群淘汰策略的小生境算法，可避免算法产生早熟现象。通过几种遗传算法的实验结果比较，证明多染色体交叉算法在多峰优化中的效果要优于传统遗传算法。

关键词: 遗传算法, 多染色体交叉, 小生境算法, 实数编码