Resonance-based bandwidth Fourier decomposition method for gearbox fault diagnosis

Abstract

This paper aims to investigate the response mechanism of faulty gears, and further proposes a resonance-based bandwidth Fourier decomposition (RBBFD) method for gearbox fault identification. To reveal the vibration patterns of the gear-pair system, a dynamic model with time-varying stiffness is established to obtain the transient response excited by local damage. The analysis results demonstrate that the impulse response of a faulty gear is essentially carried by the natural frequency instead of the widely recognized meshing frequency and its harmonies. After that, the RBBFD method is developed based on the dynamic characteristics. In the proposed method, the resonance frequencies of the gear system are estimated according to the maximum point distribution of the Fourier spectrum, and then the Fourier spectrum bandwidth optimization algorithm is introduced to decompose the vibration signal into mono-components centered on the estimated natural frequencies. A new criterion termed the characteristic frequency ratio is utilized to measure the fault-related degree of the decomposed modes, and the fault type is identified by Hilbert demodulation. The effectiveness of RBBFD is verified through two diagnostic cases. The comparison with several state-of-art works validates the superiority of the proposed method in gearbox fault diagnosis.