[3F2] Wind turbine bearing fault diagnosis by maximum generalised Gaussian cyclostationarity blind deconvolution
D Peng
North China Electric Power University, China
This study aims to assess the performance of an indicator for testing the generalised Gaussian cyclostationarity/generalised Gaussian stationary (IGGCS/GGS) as a criterion for blind deconvolution in the detection of wind turbine bearing faults. The challenges associated with diagnosing wind turbine bearing faults include two primary issues: the theoretical fault frequencies or orders may not align with the actual fault frequencies or orders; and the vibration signal deviates from a Gaussian distribution. Consequently, the application of the blind deconvolution criterion indicator of second-order cyclostationarity (ICS2), as utilised in cyclostationarity blind deconvolution (CYCBD), may encounter difficulties due to the aforementioned reasons. To address this, the criterion is modified from ICS2 to IGGCS/GGS and a tolerance band is introduced during the generation of the weighting matrix. These modifications aim to mitigate the impact of the aforementioned challenges. The effectiveness of the modified blind deconvolution approach, referred to as CYCBDβ, is evaluated in this study by comparing its performance against that of CYCBD using two distinct datasets from 2.0 MW wind turbines.
Keywords: blind deconvolution, wind turbine, bearing fault diagnosis.
Keywords: blind deconvolution, wind turbine, bearing fault diagnosis.
