Extracting the Fault Features of an Acoustic Emission Signal Based on Kurtosis and Envelope Demodulation of Wavelet Packets

This chapter presents an envelope demodulation method based on wavelet packets and kurtosis to extract the fault features of an acoustic emission signal. De-noising was performed first to reduce the noise. Then, we calculated the coefficients of wavelet packet nodes and the kurtosis value after wavelet packet decomposition. Finally, we performed an envelope spectrum analysis on the reconstructed signal based on the maximum degree of kurtosis. To correctly extract fault features of the acoustic emission signal, we compared the maximum amplitude of different nodes’ envelope spectra and the kurtosis value. The method can improve the accuracy of fault diagnosis.