Compound Fault Feature Extraction of Wind Power Gearbox Based on DRS and Improved Autogram

Abstract

Compound fault feature extraction is the key to analyzing the root cause of wind power gearbox faults. A compound fault feature extraction method based on DRS and improved Autogram is proposed. Based on the DRS method, the influence of the periodic components of vibration signals on the weak fault components is reduced. A new feature quantification index of spectral kurtosis and spectral negative entropy is designed to comprehensively evaluate the narrow-band components after maximum overlapping discrete wavelet packet transform and unbiased autocorrelation processing, so as to select the optimal filtering frequency band and accurately identify the signal components containing compound fault features. The method in this paper is applied to the compound fault diagnosis of wind power gearbox and bearing, which can effectively extract multiple fault features from vibration signals and has a good diagnostic effect.

Keywords

• wind turbine
• compound fault
• discrete random separation
• fault diagnosis
• feature extraction