遗传算法中初始群体技术的改进与实现

Improvement and implementation of technology of initial population in genetic algorithm

初始群体产生技术对遗传程序设计的进化结果有直接影响。为了改进进化结果和提高搜索效率,提出了一种递减检验的随机算法(DCRA),使其与领域经验知识相结合产生初始群体的遗传算法(EDGA)。将EDGA算法应用于生产中的圆形件排样问题,实验结果表明,该算法产生了良好的效果。DCRA和EDGA对于遗传算法的其它应用领域将有一定的指导意义。

The technology of initial population has a direct effect on the evolutional result of genetic programming. In order to improve evolutional result and searching efficiency of genetic algorithm, an algorithm named degressive checkout random algorithm （DCRA） and EDGA which hangs DCRA and experiential knowledge together to generate initial population are presented. Experiments showed good efficient when EDGA was applied in the cutting problem of circular blanks. Furthermore DCRA and EDGA can be applied in other domains.