Gated Convolutional Neural Network for Wind Turbine Blade Icing Detection

Blade icing detection plays an important role in wind turbine protection and maintenance. Employing well trained deep learning model is a promising method for blade ice detection but needs effective neural networks for sensors data analysis. In this paper, we propose a GRU-gated Convolutional Neural Network (GCNN) to better fuse the information between sensors and temporal information for icing detection. Specifically, with the superior performance of feature extraction, Convolutional Neural Network (CNN) can effectively extract the correlation information of multiple sensors data. Then GRU fuses the temporal information of the feature extracted by CNN to perform the gate of the CNN layer to control the information passed on for icing detection. The proposed method is evaluated on monitoring data generated from 25 wind turbines by two wind farms. The experimental results verify the feasibility and effectiveness of the proposed GCNN.