Fault Feature Extraction of Wind Turbine Gearbox Based on Improved Singular Value Decomposition and Noise Reduction

Abstract

Due to the complex and variable operating conditions of wind turbines, the signal-to-noise ratio is low, which is not conducive to extracting fault characteristic frequencies. Therefore, this paper proposes a method based on least squares fitting to determine the optimal rank order of singular value decomposition under low SNR, denoising the vibration signal, and then extracting the fault characteristic frequency of the wind turbine gearbox. Firstly, the original signal is reconstructed by phase space, then singular value decomposition is performed, and the singular value is obtained. Then, the order near the singular value pit(the boundary between the useful signal and the noise signal) is used as the independent variable. The singular value is used as the dependent variable, and the least squares fitting is performed in turn, and the fitting error is obtained. Finally, the optimal rank order is determined by the order corresponding to the minimum fitting error, and the signal noise reduction is realized, and the gearbox is faulty feature extraction. This method is verified by the actual gearbox fault data. The experimental results show that this method can not only improve the signal-to-noise ratio, but also facilitate the extraction of fault characteristic frequencies.