An Effective Chatter Detection Method in Milling Process Using Morphological Empirical Wavelet Transform.

The empirical wavelet transform (EWT) has shown its effectiveness in some applications. However, when noisy and nonstationary signals are analyzed, some local maxima may appear and be retained in the peak sequence mistakenly, so improper segmentation in the frequency domain will occur. In our research, the morphological EWT (MEWT) method is proposed based on morphological filters (MFs) and the 1-D Otsu method to mitigate the boundary segmentation drawback of the EWT, and it can be applied in chatter detection because of its good performance in finding the optimal chatter frequency band. First, the local maxima distribution plane is calculated using MFs. Then, the 1-D Otsu method processes the distribution plane and then achieves the optimal threshold for MFs. For MEWT’s better application in chatter detection, an automatic selection of the parameter $K$ using kurtosis is proposed and it breaks away from the previous empirical selection. Finally, since it is known that energy will transfer to chatter frequency bands when chatter occurs in the milling, the energy entropy of each subsignal used is to detect chatter frequency bands and choose the optimal chatter monitor indicator. The relevant simulation and experimental signals have been analyzed for verifying the validity of the MEWT and the novel chatter detection method with strong sensitivity to chatter. [ABSTRACT FROM AUTHOR]