The fault diagnosis of rolling bearing in gearbox of wind turbines based on second generation wavelet

Abstract

There are two difficulties when we try to diagnose the fault of rolling bearing in gearbox of wind turbines. Firstly, gearbox typically has multi-level structure and every level produces different speed ratios. The complex structure increases the difficulty of identifying accurately faults of rolling bearing on intermediate shaft. Secondly, it's not easy to determine fault feature frequency components in spectrum. Currently the research of gearbox fault diagnosis is based on the vibration signal collected from box. Obviously there is interference on the signal because the procedure of transmission consists of too many links and sensor has various working conditions. This paper aims at solving the two problems. The first part of this paper mainly deals with rotating speed and speed ratio of shafts in gearbox and the later part processes vibration signal. Wavelet is a widely-used analysis method for vibration signal. Unlike traditional wavelet, second-generation wavelet is independent of Fourier transform and owns self-adaption. It is easier to detect the fault characteristic frequency in noisy signal. Firstly, the original signals are decomposed and reconstructed by second-generation wavelet. Then, the components with the most fault information are selected for Hilbert envelope spectrum analysis according to the parameter Kurtosis. Finally, the fault characteristic frequency is extracted based on the Hilbert envelope spectrum and the fault type is also identified. The result shows that this method is correct and efficient.