eyeDNA : Tool Condition Monitoring for a desktop CNC milling machine

Abstract

Tool wear is a major obstacle to realizing full automation in metal cutting operations. In this thesis, we designed and implemented a low cost Tool Condition Monitoring (TCM) system using off-the-shelf sensors and data acquisition methods . Peripheral end milling tests were done on a low carbon steel workpiece and the spindle vibration, cutting zone temperature and spindle motor current were recorded. Features from these data sources were used to train decision tree models in MATLAB with the aim of classifying the stages of tool wear. Results showed that the feature sets fusing information from all data sources performed the best, classifying the tool wear stage with up to 93% average accuracy.