An analysis method of gear fault diagnosis based on Chirp-Z transform and local mean decomposition

Abstract

A combined fault diagnosis method based on the CZT (Chirp-Z transform) and LMD (Local mean decomposition) marginal spectrum is proposed in consideration of the complex nonlinearities and non-stationary properties presented by the fault vibration signal in the gear. In this method, the original vibration signal is initially decomposed through LMD into several stationary PF (product function) components with physical meaning, and then the PF components that contain main fault information are selected for CZT local zoom analysis. The LMD marginal spectrum is determined by calculation of instantaneous amplitude and frequency information of each PF to characterize the overall distribution of fault signals. The results of the experiment research on the simulated signals and the gear crack faults indicate that with high fault recognition rate, the combined fault diagnosis method can be adopted to reflect the characteristics in variation and distribution of fault signals both from local and overall perspectives.