基于奥运选拔机制的超种群遗传算法

遗传算法有很多优点,但早熟问题有待解决。本文在对遗传算法早熟问题的机理进行深入分析的基础上,基于奥运会优秀运动员选拔机制,提出了一种用多种群进行搜索,然后构建一个新种群的超种群遗传算法。该算法可以避免传统遗传算法中因为模式单调化而造成的早熟。数值仿真表明,该方法能以较大的概率收敛于全局最优解。

The Super Population Genetic Algorithm base on the choosing mechanism of the Olympic Gmes

The Genetic Algorithm has a lot of advantages, but the premature convergence problem remains to solve. This thesis that is on the basis of analyzing the reason of the premature convergence in depth and the choosing mechanism of the Olympic Games proposes the Super Population Genetic Algorithm(SPGA). SPGA adopts several populations to search, and then build a new super population to evolve. The SPGA can avoid the premature convergence that is produced by lack of patterns in the traditional Genetic Algorithm. Numerical examples show that SPC, A can converge to the global optimal value with high probability.