Identification of wind turbine gearbox weak compound fault based on optimal empirical wavelet transform

Abstract

A double modulation phenomenon is found in wind turbine gearbox compound fault signals, consisting of a resonance modulation frequency band and an asymmetric modulation frequency band. The modal aliasing and the double modulation phenomena make it difficult to use empirical wavelet transform to obtain the spectral components in the meshing modulation regions. Our main contribution is to establish an optimal empirical wavelet transform (OEWT) framework for weak feature extraction of compound faults, which incorporates the optimal velocity synchronous linear chirplet transform (OVSLCT) and the simulated annealing (SA) algorithm to obtain the optimal parameters of the filter bank. The frequency boundary of the fault component is obtained through OVSLCT rather than the extreme value of the spectrum. A filter transition band width optimization scheme is proposed in which the optimal transition band width of the filter bank in EWT is optimized by the SA algorithm. Compared with the original EWT and variational mode decomposition, OEWT can better obtain the compound fault characteristic frequency and solve the mode aliasing problem.