Data-driven fault identification of ageing wind turbine

This paper proposes an ageing evaluation method for wind turbine system by using a data driven method. This method directly uses the input and output data of the wind turbine system, and the autoregressive with exogenous (ARX) model to identify the wind turbine system. The input and output data include wind speed, generated power, and pitch angle, and they are generated by a wind turbine simulation model with four ageing cases: mechanical power, magnetizing inductance, pitch angle controller gain and pitch angle change rate. By using the generated power and pitch angle data of wind turbine under different ageing levels, the data-driven models can be obtained. By comparing the model parameters in different states identified by the ARX model, results show that the degree of ageing can be reflected by the parameter changes. This demonstrates that the method can detect the ageing of wind turbines.