Abstract: It is pointed that wavelet analysis has powerful ability for weak signal detection, which helps it to be used for detection and fault diagnosis of acoustic emission signals well. However, in wavelet analysis of fault acoustic emission(AE) signals nowadays, the diagnosis results using the general wavelet functions are not the best, and some new wavelet functions need to be created at once. By analyzing the characteristics of typical AE signals initiated by mechanical faults or damages, and according to the extracting principle of fault characters of AE signals and the construction method of continuous wavelet function, a wavelet function is designed. When applying the function and Daubechies wavelet for fault diagnosis of rolling bearings based on AE technique at the same time, the results using the former are more clear, accurate and reliable. Both theory analysis and experiment research prove that the wavelet function is scientific and effective.

Key words: Acoustic emission, Fault diagnosis, Feature extraction, Rolling bearing, Wavelet function, Wavelet transform