Encoder-based angular-domain weighted allocation deconvolution for fault diagnosis of gearboxes under speed-varying conditions

The fault diagnosis of a planetary gearbox under the speed-varying condition without another speed information is always challenging, especially when the vibration signal has a low signal-to-noise ratio (SNR). To solve the problem, a novel angular-domain weighted allocation deconvolution (AWAD) method based on the encoder signal is proposed in this paper. First, the rotational speed information and fault-related information are simultaneously presented as different patterns of the single encoder signal, which provides the ideal tool for the non-stationary problem. Second, benefiting from the advantage of the encoder signal, the prominent filtering performance of deconvolution method (DM) is highlighted, and a new angular-domain DM is designed under the speed-varying condition. Meanwhile, an angular-domain weighted allocation strategy based on the second-order cyclostationarity indicator is introduced to further enhance the weak fault feature. Finally, the superiority of the proposed AWAD is demonstrated to adaptively and accurately extract the weak fault feature as well as be robust to the low SNR and speed-varying condition without any prior knowledge using simulated and experimental data collected from the encoder signal with a complex gearbox fault.