Fault Extraction of Wind Turbine Rolling Bearings Using FDEO and the Improved ACYCBD

Abstract

Rolling bearings normally work under variable speed and load conditions, and their fault becomes hard to be extracted. In this study, for a method of extracting rolling bearing fault is proposed based on the frequency domain energy operator (FDEO) and an improved adaptive maximum second order cyclostationarity blind deconvolution (ACYCBD). Firstly, high speed shaft means the frequency provided by the SCADA system is used to find the optimal vibration component for further analysis. The instantaneous frequency of the vibration component is then estimated using FDEO for order tracking. Finally, for the purpose of addressing the problem such as multiple vibration components existing in a vibration signal collected from wind turbines, which may mask the vibration component of interest, the improved ACYCBD is then used to extract the fault feature. The industrial results show that the proposed method is able to extract the faulty feature at an early stage without the interference of other vibration sources.

Keywords

• Wind turbine;Rolling bearing fault;Frequency domain energy operator;Adaptive maximum second order cyclostationarity blind deconvolution