Axlebox bearing fault diagnosis method for rolling stock combining improved CEEMD and MOMEDA

To enhance the effectiveness of diagnosing axial bearing fault signals in moving trains, this study proposes a method that combines improved complementary set empirical modal decomposition and optimal minimum entropy deconvolutional adjustment. There are plans to develop a screening method based on intrinsic modal function components to further boost the diagnostic procedure's effectiveness. The simulation experimental validation showed that the fault eigenfrequencies from 1 to 7 octave may be identified by the research-proposed method after envelope spectral analysis. Case Western Reserve University dataset validation indicated that the proposed method is superior in terms of bearing fault signal processing results. The time-domain amplitude of the inner ring fault signal increased by 50% and was increased at all times compared to other methods. The eigen frequency of the inner ring fault signal was found to be between 1 and 9 octaves, whereas the outer ring fault signal was found to be between 1 and 14 octaves. The findings show that the suggested approach is capable of accurately diagnosing axlebox bearing fault signals in the locomotive group and of directly localizing the fault location based on the envelope spectrogram's characteristic frequency.