Vibration based Assessment of Tool Wear in Hard Turning using Wavelet Packet Transform and Neural Networks

Demanding high dimensional accuracy of finished work pieces and reducing the scrap and production cost, call for devising reliable tool condition monitoring system in machining processes. In this paper, a tool wear monitoring system for tool state evaluation during hard turning of AISI D2 is proposed. The method is based on the use of wavelet packet transform for extracting features from vibration signals, followed by neural network for associating the root mean square values of extracted features with tool flank wear values of the cutting tool. From the result of performed experiments, coefficient of determination and root mean square error for the proposed tool wear monitoring system were found to be 99% and 0.0104 respectively. The experimental results show that wavelet packet transform of vibration signals obtained from the cutting tool has high accuracy in tool wear monitoring. Furthermore, the proposed neural network has the acceptable ability in generalizing the system characteristics by predicting values close to the actual measured ones even for the cutting conditions not encountered in the training stage.