Generalized Demodulation Method Based on Multi-scale Chirplet and Sparse Signal Decomposition and Its Application to Gear Fault Diagnosis

Abstract

On the basis of multi-scale chirplet and sparse signal decomposition, a new method of generalized demodulation is proposed to be applied to the fault diagnosis of gear in a state of non-stationary rotating speed. The method of generalized demodulation can transform multi-component and non-stationary signals with time-frequency distribution consisting of curves into stationary signals with time-frequency distribution consisting of linear lines which are parallel to the time-axis. So after the implementation of generalized demodulation, the signals satisfy the standard of stationary demand in FFT analysis. The key point in generalized demodulation is how to obtain the phase functions of the multi-component signals. Through the sparse signal decomposition based on multi-scale chirplet of the signal, phase functions of the component signals are acquired and then generalized demodulation and frequency analysis are adopted to get the fault feature frequency components. The proposed method is suitable to analyze the multi-component AM - FM signals which fluctuate with the rotating speed. A simulation example and an example of its application prove the effectiveness of the method applied to fault diagnosis of the gearbox with rotating speed fluctuation.