支持多任务集中下料的零件分组优化方法

Grouping optimization method of large-scale parts supporting centralized cutting stock

在多任务集中下料中,针对算法在处理大规模零件下料问题时易陷入时间效率和材料利用率矛盾的问题,提出基于下料零件样本的分组优化方法。该方法首先利用零件样本完成大部分待下料零件的过滤分组,将形成的零件球体中心作为初始聚类中心,对余下的零件进行硬C均值聚类;然后基于零件下料配合特征将零件相似组重组为若干个分组,在优化中根据各组材料利用率对零件的组间分布进行动态修正,合并各组优化结果,得到原问题的下料方案。基于该方法开发了一种条材优化下料系统,实际应用表明,相对于没有零件样本参与的随机分组优化,该方法在处理多任务集中下料时可以获得更优更稳定的下料方案。

To resolve the contradiction of time efficiency and material utilization ratio in Centralized Cutting Stock Problem（CCSP）,a grouping optimization method based on parts samples was proposed.Based on the guiding role of parts samples,filtration grouping of most parts was accomplished.The formed parts group center was taken as its initial clustering center,and the remaining parts were completed by Hard C-Means（HCM）.According to cutting stock combination characteristics of parts,the parts groups were recombined into several parts clusters.A dynamic compensation strategy based on parts clusters’ material utilization ratio was adopted to adjust parts in different clusters.Combined each optimization result,the cutting plan of CCSP was obtained.Based on proposed method,a cutting system of bar stock was proposed,and compared with random grouping optimization without parts samples,the proposed method could obtain better cutting plan.