基于模型的子空间聚类与时间段蚁群算法的合同生产批量调度方法  

Order Production Scheduling Method Based on Subspace Clustering Mixed Model and Time-section Ant Colony Algorithm

作　　者：王利[1] 高宪文[1] 王伟 王琦[1] 

机构地区：[1]东北大学信息科学与工程学院,沈阳110004 [2]大连理工大学电子信息与电气工程学部,大连116024

出　　处：《自动化学报》2014年第9期1991-1997,共7页Acta Automatica Sinica

基　　金：国家自然科学基金(61034003;61104157)资助~~

摘　　要：针对目前冷轧薄板厂生产流程复杂、大量的多品种小批量合同并线生产,导致难以制定生产计划的问题,本文提出了混合模型子空间聚类(Subspace clustering mixed model,SCMM)方法,以合同中待加工钢卷的宽度、冷轧机组的入口厚度、出口厚度以及合同的交货期为约束,对待生产合同进行组批.依据冷轧厂实际生产过程,将冷轧机组视为核心节点,考虑准时交货、在制品库存和生产流向产能分配的要求,对组批后的生产合同建立全流程合同计划模型,并且利用提出的时间段蚁群算法(Time-section ant colony optimization,TSA),制定合同计划.利用生产过程的实际数据测试,本文的方法优于人工排产,可以满足制定冷轧薄板全流程生产计划的要求.Because of complexity of production processes, a rational production plan could not be established if a large amount of small batch orders are to be produced in a whole flow line in the original sequence of orders. A subspace clustering mixed model （SCMM） is established which is used to group the orders. The constraints of the grouping method are the width of steel roll, thickness of entrance, exit of tandem cold mill, and delivery date. According to the production flow in BaoSteel Cold Rolled Sheet Mill, the tandem cold mill is a bottle neck of the enterprise. For realization of the order planning of the whole flow, a planning model and the time-section ant colony optimization （TSA） are proposed based on the delivery date, inventory and production route of products. Experiments with real data of production processes indicate that the proposed methods are superior to human scheduling and meet the requirement of the order planning.

关 键 词：冷轧 子空间聚类 蚁群算法 生产计划与调度 

分 类 号：TG338[金属学及工艺—金属压力加工] TP18[自动化与计算机技术—控制理论与控制工程]