The Health Status Prediction of the Wind Turbine Based on the Anomaly Analysis and the LSTM Prediction

Abstract

The traditional data driven faulty prediction model aims to build a nonlinear mapping between the reference signal input and the target degradation index. This kind of faulty detection model has achieved a certain degree of detection accuracy, However, most of these prediction models execute prediction based on the current input data [1]. With the development of the modern industrial process, the modern industrial manufacturing equipment is becoming high complexity and large scale. The health situation of the manufacturing equipment is usually associated with not only the current input data, but also the historical data. In this article, the hybrid approach combined with the anomaly clustering and the LSTM is proposed for the faulty classification of the wind turbine. The anomaly analysis is first used to choose the input sensing signal which can well represent the situation of the health situation of the wind turbine, afterwards, the chosen signal input is used as the input of the LSTM data. The result shows that the proposed model performs quite well in the faulty prediction of the wind turbine.