Wind turbine pitch bearing fault detection with Bayesian augmented temporal convolutional networks

There are few studies on the fault diagnosis of deep learning in real large-scale bearings, such as wind turbine pitch bearings. We present a novel method, Bayesian augmented temporal convolutional network (BATCN), to filter raw signal bearing defect detection. This which employs neural networks, is designed capture dependencies signal, with focus non-stationary relationships collected signals. By referring thoughts optimization, our approach can spontaneously find best patch length that influences extraction during filtering process, avoiding manual tuning this hyper-parameter. BATCN method first performed simulation signals and an open-source dataset general then validated industrial bearings both lab farm, where have been operated for over 15 years. The results show work well slow-speed