Milling Tool Condition Monitoring Based on an Integrated Wireless Vibration Sensing Tool Holder

Abstract

Tool condition monitoring (TCM) is crucial for smart manufacturing and cutting vibration signal is proven to be highly related to tool wear state. In this paper, a wireless smart tool holder is designed for online vibration signal sensing for TCM with accelerometer embedded close to vibration source and signal processing circuits integrated, showing good performance of vibration sensing ability compared with traditional wired ways. Cutting experiments are designed with cutting parameters of great varied range to guarantee the generalization ability of TCM algorithm for different machining conditions and vibration signal of whole tool life cycle is collected by smart handle. Then feature extraction and selection are studied to provide valuable information and artificial neural network algorithm is realized. Results show the algorithm has an accuracy of 85.0% with poor performance in distinguishing some wear states. To solve this problem, an optimized method based on two ANNs in series with new feature sets is proposed. The optimized algorithm has an accuracy of 90.0% with an accuracy increase of 16.8% and the average predicted probability increase of 15.0% in initial wear samples. In spite of speed sacrifice, the optimized algorithm makes progress in recognition accuracy and data confidence level.