Hierarchical cluster analysis for pattern recognition of process conditions in die sinking EDM process monitoring

Abstract

Die sinking EDM processes require continuous monitoring due to the typically severe application requirements, especially in advanced aerospace parts machining, where part quality and machining time are main concerns. As the process conditions cannot be recognized based on the behaviour of a single monitored value, it is necessary to consider a number of relevant sensor signals together. The aim of this research work is to recognize the machining conditions which lead to an improper process performance, e.g. by increasing machining time and causing unacceptable part quality, and to highlight the most relevant sensorial features. Using the Real Time Acquisition (RTAQ) module installed on an AgieCharmilles FORM P 600 sinker spark erosion machine, eight process parameters are acquired. Hierarchical cluster analysis is then applied to identify different groups of improper process conditions based on relevant features extracted from the EDM process parameters.