Typical Machining Process Mining of Servo Valve Parts Based on Self-adaptive Affinity Propagation Clustering

Abstract

The effective reuse of the typical process is a key factor in the efficiency of process planning. Therefore, the typical process mining was studied. First, the current situation and problems in process planning were analyzed. Modeled by attributed directed graph, the process was represented in mathematical expression systematically. And to measure the similarity between processes, the similarity between processes cells and the similarity between process routes were defined. Self-adaptive affinity propagation was used in the clustering of typical process and effective clustering results were evaluated by the Silhouette index. Finally, the mining method was used in discovering typical process from machining process of 3 sets of servo valve sleeves in 5 different models and proven to be effective.