采用遗传算法对成组技术中的零件进行近优分蔟，提出了用０、１码分段构造染色体和分段交叉、变异的策略来满足Ｐ－中位模型复杂约束的要求。实验结果表明，当以类内样本距离之和作为评价准则时，该算法明显好于Ｋ－平均算法。 The genetic algorithm is used to approximate the components in the group technique. The strategy of constructing the chromosomes and segmentation crossover and mutation by using 0 and 1 yards is proposed to meet the requirements of the complex constraints of the P-median model. The experimental results show that the algorithm is obviously better than the K-means algorithm when the sum of the distance of the samples in the class is used as the evaluation criterion.