基于遗传算法与DSM的产品结构分解聚类方法

产品结构的分解聚类在产品开发中有着重要的地位,在用设计结构矩阵(D es ign structure m atrix,DSM)对产品结构进行建模的基础上,通过遗传算法实现了产品结构分解聚类的智能化和分解聚类结果的最优化。在算法的设计过程中开发了一种对DSM进行二维编码的方法,并给出了在二维编码基础上的多点杂交和基本变异方法。在构造适应度函数时,综合考虑了DSM模型中各元素之间的联系、聚类的数目以及各聚类中元素的数目。最后以某摩托车发动机为例,用该算法实现了产品结构DSM模型的智能化分解聚类,验证了该算法的可行性。

Decomposition and Clustering of Product Architecture Based on Genetic Algorithms and DSM

Decomposition and clustering of the product architecture is important during product development.Based on building the product architecture with the design structure matrix(DSM),decomposition and clustering of the product architecture is intelligentizedly realized by genetic algorithms(GAS).During designing the algorithms,2-D coding about DSM model is presented,and the corresponding multi-point crossover and basic mutation are introduced.When designing the fitness function,the relations between elements in DSM,the number of clusters,and the number of elements in each cluster are considered.(Finally),the intelligentized decomposition and clustering with algorithms are applied to an engineering example.The result validates the algorithms.