A gearbox fault diagnosis method based on MKurt spectrum and CYCBD

Maximum second-order cydostationary blind deconvolution (CYCBD) is a good signal denoising method, which can be employed for gear fault diagnosis. However, CYCBD is highly dependent on the pre-period of the measured signal, which needs to be appropriately predetermined. To address this issue, a new method based on the multi-point kurtosis (MKurt) spectrum and CYCBD is proposed, which can be used for extracting fault features in a gearbox. First, deconvolution period T the key parameter of CYCBD, is accurately selected using the MKurt spectrum. Then, according to the selected key parameter, CYCBD is employed to process the gearbox signals and identify the fault type. This proposed method is applied to the analysis of single and compound fault signals of a gearbox, the large gear and pinion fault signals under strong background noise are separated. Finally, the fault signals obtained by CYCBD are analysed using an envelope spectrum to extract the fault features. The two case studies demonstrate that the proposed method can effectively identify the gear faults. Moreover, the results show the superior effectiveness and reliability of this proposed method compared with the maximum correlation kurtosis deconvolution (MCKD) method.