Feature Extraction of Gearbox Early Fault based on ISGMD and MED

Abstract

Aiming at the difficulty in identifying early faults and compound faults of gearboxes under strong noise background，a method of extracting fault features based on the combination of improved symplectic geometry mode decomposition （ISGMD） and minimum entropy deconvolution （MED） is proposed. Firstly，the signal is preprocessed by minimum entropy deconvolution to highlight the fault impact component in the signal. Then，the fault enhancement signal is adaptively decomposed into several symplectic geometric components through improved symplectic geometric modal decomposition，and the sensitive symplectic geometric component with the largest kurtosis value is selected according to the maximum kurtosis criterion. Finally，an envelope analysis of the selected sensitive symplectic geometric components can effectively extract the fault features of the gearbox. The effectiveness of the method is verified by experimental.

Keywords

• Gearbox;Symplectic geometry mode decomposition;Minimum entropy deconvolution;Early fault;Feature extraction