Wind turbine gearbox fault diagnosis based on wavelet domain stationary subspace analysis

Fault-related signals of wind turbine gearbox are non-stationary, transient and weak, which are often mixed together with gear meshing signals and submerged in background noise. A new wind turbine gearbox fault diagnosis method based on continuous wavelet transform (CWT) and stationary subspace analysis (SSA) is presented. The SSA is a blind source separation technique that can extract stationary and non-stationary source components from multi-dimensional signals without the need for independency and prior information of the source signals. Multi-scale analysis ability inherent in CWT allows for decomposing one dimensional signal into multi-dimensional signals, which can be naturally used as inputs to SSA to obtain the stationary parts and non-stationary parts of the original signal. Subsequently, the selected non-stationary component is analyzed by the envelope spectrum to identify potential fault-related characteristic frequency. Experimental studies from a real wind turbine gearbox test have verified the effectiveness of the presented method.