Gear fault diagnosis via demodulation analysis based on NA-MEMD

Abstract To address the non-stationary characteristics of gear vibration signals under fault conditions, a noise-assisted multivariate empirical mode decomposition (NA-MEMD) approach is employed for gear fault detection. This study presents a fault diagnosis methodology combining NA-MEMD with envelope demodulation analysis. The NA-MEMD technique adaptively decomposes vibration signals into multiple mono-component intrinsic mode functions. Fault-related components are subsequently subjected to envelope demodulation processing to extract characteristic fault signatures from gear vibration data. Experimental validation using actual gearbox fault cases demonstrates that the developed approach successfully identifies gear defect features through effective signal decomposition and demodulation analysis. Results confirm the method’s capability to enhance fault feature extraction for mechanical transmission systems.