Abstract: Aiming at the problem of quick process parameters set for new product with complex process which responded to the necessity of customer in process manufacturing environment,a data-driven matching method for processing parameters was proposed as process knowledge service.The features extracted from different products were acquired in manufacturing data by information theory,and the relationship model of product standards and process parameters based on ensemble-classification was constructed to realize the process similarity analysis.Through comparing the similarity of historical production data and target product's manufacturing requirement,the near-optimal process parameters were selected.The proposed method was applied to the hot-rolling process of steel plate,and the result showed the effectiveness on providing process knowledge service and conformity for dealing with coupling knowledge discovery.

Key words: process manufacturing industry, process parameters matching, data-driven, knowledge discovery, similarity analysis, ensemble-classification, information theory

摘要： 针对流程制造业在个性化制造环境中工艺参数难以快速决策以响应用户需求的问题,提出一种基于数据驱动的流程工艺参数匹配方法。从海量制造工艺数据中选取相近工艺的产品生产数据,基于信息论从制造工艺参数中提取能够划分不同产品的工艺特征,基于集成分类构建产品制造要求与工艺特征的关系模型,并建立产品工艺相似度函数。通过对比历史生产数据与目标产品制造要求的相似度实现工艺参数匹配。将该方法应用于钢板热轧环节,验证了所提方法可以有效提供工艺知识服务,并能够处理耦合性特征的知识发现。

关键词: 流程制造业, 工艺参数匹配, 数据驱动, 知识发现, 相似度分析, 集成分类, 信息论