Bearing fault diagnosis using wavelet analysis

The development of real-time fault detection and identification systems will allow a migration from expensive scheduled based maintenance to the more efficient, less costly alternatives of condition based maintenance. This paper presents a detailed analysis of different methods using wavelet analysis in the diagnosis of faults in rotating machinery from its mechanical vibrations. More specifically, the discrete wavelet transform and the discrete wavelet packet transform will be discussed, as well as the choice of the mother wavelet and the different features that can be extracted from such transforms.