Application of Artificial neural network and wavelet packet transform for vibration signal based monitoring in mechanical micro drilling

In order to achieve high quality and productivity in microdrilling, monitoring of the prefailure phase and detection of tool breakage is very important. In the present work, vibration signals have been studied during micro drilling operations to monitor the prefailure phase of the micro-drills. These signals have been processed in time domain and time-frequency domain to extract tool wear sensitive features. An Artificial neural network (ANN) has been developed from time domain feature and wavelet packet features of vibration signals to predict the hole number of the micro-drilling at different spindle speed and feed. The prediction of drilled hole number using ANN model is in good agreement to experimentally obtained drilled hole number. It has been found that wavelet packet feature based ANN model outperforms the time domain feature based ANN model.