Gearbox fault diagnosis using independent component analysis in the frequency domain and wavelet filtering
Xinhao Tian, Jing Lin, Ken R. Fyfe, Ming Jian Zuo · 2003 IEEE International Conference on Acoustics, Speech, and Signal Processing, 2003. Proceedings. (ICASSP '03). · 2003
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, /spl beta/, 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.