基于均匀分割的多种群并行遗传算法

针对标准遗传算法在处理多峰函数优化问题时易出现的成熟前收敛现象，在讨论了模式样本分布特点的基础上，提出了一种通过均匀分割对种群分类的多种群并行遗传算法。由均匀分割可以得到几个既不重叠又都能反映函数整体性质的子空间，在这些子空间上并行搜索最优解，同时将每一代在各自空间上搜索到的优秀个体集中在一起，进而在全空间上搜索最优解的具体位置。由于在这些子空间上的搜索是彼此独立的，所以同时发生“早熟”现象的机会大大降低。理论分析和对多峰函数的仿真结果均表明，该算法在不影响收敛速度的条件下，发生成熟前收敛的概率明显下降。

Multipopulation Parallel Genetic Algorithm Based on Even Partition

Through discussing some distributive characters of schemata samples, a multipopulation parallel genetic algorithm based on even partition is presented. The method solves the problem to premature convergence of standard genetic algorithm dealing with multimodel functions. Several subspaces which have no overlaps and reflect whole characters of the functions are gained through even partition. This genetic algorithm realizes parallel optimum search in each subspace and collects excellent individuals after every evolution, then searches the optimum solution in whole space. Because the searches in these subspaces are independence, the possibility of occurring premature at the same time falls greatly. Without affecting convergence rate, theory analysis and experimental results show that the method is very effective.