18 Self-Adapting Envelope Analysis Researches on Rolling Element Bearing

The vibration signal from the initial fault of rolling element bearing is characterized by short time impacts, which is weak in energy and buried in the surrounding noise. This feature makes traditional envelope analysis method often disturbed in application. In this paper, a scheme is introduced, which uses autoregressive model preprocessing the failure data and thereby relatively enhances the transient impact in the failure signal. Subsequently, the self-adapting complex Morlet wavelets envelope analysis method based on maximizing kurtosis criterion is done, then, a good transient impact is extracted from the preprocessed signal. This approach avoids the deficiency of traditional envelope needing to set centre frequency and envelope bandwidth manually. Experiments verify the feasibility and effectiveness of the introduced scheme.