A Fault Diagnosis Model for Wind Turbine Blade Using a Deep Learning Method

A wind turbine plays a pivot role in the field of energy supply. A stable working state of the wind turbine is highlighted in power generation. For this reason, it is thus of great interest to develop a fault diagnosis method for status identification of wind turbine. In this work, the imbalance of turbine blade is investigated on the task of fault diagnosis. Vibration sensors are attached to the blade to detect vibration signals. A deep learning based method using gated convolution neural network and long short term memory unit, together with the attention mechanism, is established. The application of the proposed model is capable of capturing the most-related features that characterize the working states. The working performance of the proposed model is validated on the samples from various working conditions. The experimental results set solid evidence of a high accuracy.