Abstract: Aiming at the problems of enterprise process data and knowledge mining, a method of extracting product typical process routs based on intelligent clustering analysis was presented. A similarity factor between two process routs was established and a multi-level comprehensive measurement method for calculating the similarity between two process routs was proposed. Based on the similarity calculation, a process rout design structure matrix was constructed and the noise reduction processing was applied to the matrix data. To reduce the difficulty and complexity of clustering division, the particle swarm optimization was used to realize the intelligent clustering division of process rout design structure matrix. The typical process routs were extracted from the clustering clusters consequently. A mechanical press enterprise was taken as an example to extract typical process routs from the process data, and the effectiveness of proposed method was verified.

Key words: process similarity, design structure matrix, intelligent clustering, particle swarm optimization, typical process rout

摘要： 针对企业工艺数据与知识挖掘问题，提出应用智能聚类分析技术提取产品典型工艺路线的方法。构建了工艺路线的相似度度量因子，提出了对工艺路线进行相似度计算的多级相似度综合度量方法，在相似度计算基础上，构建了工艺路线设计结构矩阵,并对矩阵数据进行降噪处理；为降低聚类划分的难度和复杂性，运用粒子群优化算法实现了工艺路线设计结构矩阵的智能聚类划分，并从聚类簇中提取到典型工艺路线。以机械压力机企业工艺数据的典型工艺路线提取为例，验证了该方法的有效性。

关键词: 工艺相似度, 设计结构矩阵, 智能聚类, 粒子群算法, 典型工艺路线