Autocorrelated Envelopes Assisted Adaptive MED for Gearbox Rolling Bearing Fault Diagnosis

Effective fault characteristic extraction is crucial for the diagnosis of gearbox bearing. Minimum entropy deconvolution (MED) is a popular method for fault feature extraction. However, the deconvolution results are sensitive to the choice of inverse filter parameters. To overcome this limitation, an adaptive minimum entropy deconvolution (AMED) is proposed. AMED searches for the optimal filter length when the energy proportion of the fault-related features in the vibration signal is highest. In addition, this paper explores the use of autocorrelation of the envelope signal to further strengthen the fault characteristic frequencies in the signal. The properties of autocorrelated envelopes is integrated with the AMED approach to develop a novel weak bearing fault diagnosis method. Simulation and experimental results validate that the proposed method is effective in enhancing the weak features of the bearing associated with the failure, and a comparison with several existing methods demonstrates the advantages of the proposed method.