In this paper, we combine independent component analysis in the frequency domain (ICA-FD) and Morlet wavelet filtering for gearbox fault diagnosis. Collected vibration signals from a gearbox are separated into two components with ICA-FD. Morlet wavelet filtering is then applied to the separated components. The optimal shape parameter β of the basic Morlet wavelet is obtained by minimizing the wavelet entropy. Better diagnosis results are obtained with this combination than using wavelet filtering alone.
展开▼