Abstract:  Traditional machining processes of mould parts which were relied on manual work had low efficiency and long design cycle， and might not adapt the needs of efficient and intelligent productions of modern mould parts with multi-variety and small batch， and thus a machining process decision-making model was established based on case reasoning. A method of calculating the weight of feature attributes of mould parts was proposed by the rough set theory； the numerical and character local similarities were accurately characterized by adopting the method of normalization and cosine similarity； the similarity threshold of system was dynamically obtained by combining the static and dynamic similarity threshold； based on the process decision-making algorithm combined with K-means clustering and K-nearest neighbor(KNN)， the process files might be generated quickly， accurately， and intelligently. The enterprise production testing results show that using the constructed machining process decision-making model of mould parts， the time for analysis and process route formulation of mould parts may be decreased by 64.5%， the time for machining process preparation of mould parts may be decreased by 40%， the proportion of unreasonable process design may be reduced by 82.8%， and the proportion of missing procedure machining may be reduced by 69.6%， which greatly improve the efficiency of process design and the machining quality of mould parts. 

Key words:  , similarity, mould part, case reasoning, machining process design, decision-making model

摘要： 针对传统模具零件加工工艺依靠人工编制而导致的效率低、设计周期长，无法适应现代模具零件多品种、小批量、高效智能化生产的需求，构建了一种基于实例推理的模具零件加工工艺决策模型。采用粗糙集理论，提出了模具零件特征属性权值计算方法；采用归一化和余弦相似度方法，对数值型和字符型局部相似度进行了准确表征；基于静态及动态相似度阈值相结合的方法，动态获取了系统相似度阈值；提出K-means聚类算法与最近邻（KNN）算法相结合的工艺决策算法，实现了加工工艺文件快速准确智能化生成。企业生产测试结果表明，采用所构建的模具零件加工工艺决策模型，可使得零件模型分析及工艺路线拟定时间缩短64.5%，模具零件加工工艺编制时间缩短40%，工艺设计不合理比例减小82.8%，漏工序加工比例减小69.6%，极大地提高了工艺设计效率及模具零件加工质量。

关键词: 相似度, 模具零件, 实例推理, 加工工艺设计, 决策模型