Radial Basis Function Network Based Monitoring of Tool Wear States

In this paper, combination of wavelet packet decomposition (WPD) and neural networks (NN) was used to identification the experimental cutting torque data of drilling operations previously. It consists of three steps: firstly, decomposition cutting torque from the original signals by WPD; secondly, extracting wavelet coefficients of different wear states (i.e., slight, normal, or severe wear) with signal features adapting to Welch spectrum; finally, the spectrum feature vectors identify by using the radial basis function neural network (RBFNN). The experiments on different tool wears states of monitoring and identification are significant and effective.