A new feature acquisition method for indirect drill wear monitoring

抄録

The methodologies or technologies applied for indirect monitoring of machining processes can be summarized as sensor/sensor system, signal processing, feature generation, feature extraction, feature selection and decision making. This paper concerns the feature generation, feature extraction and feature selection methods in the monitoring of drilling. The features are generated with forces converted from thrust force and torque, extracted by wavelet packet transform (WPT) and selected using principal component analysis (PCA). And the acquired features are then used as inputs to a back-propagation neural network (BPNN) to predict the drill corner wear.