Abstract: Based on the definition of the non-component signal, namely intrinsic scale component (ISC) whose instantaneous frequencies own physical sense, a newly self-adaptive signal decomposition method, the local characteristic-scale decomposition (LCD) is proposed. LCD can decompose any complicated signal into a number of ISC whose instantaneous frequencies own physical sense and is especially suitable for processing the multi-component amplitude-modulated and frequency-modulated signal. Since the gear fault vibration signal is multi-component amplitude-modulated and frequency-modulated signal, the local characteristic-scale decomposition method is especially applicable for the gear fault diagnosis. The LCD method is compared with the empirical mode decomposition (EMD) and the local mean decomposition (LMD) method; then the results show the superiority of the LCD method. Simultaneously, aiming at the modulated features of the gear fault vibration signal, the LCD is combined with the envelope analysis method and applied to the gear fault diagnosis. The analysis results from the actual gear fault vibration signal demonstrate that the LCD method can be applied to the gear fault diagnosis effectively.

Key words: Gear fault diagnosis, Intrinsic scale component, Local characteristic-scale decomposition, Modulation, Self-adaptive signal decomposition