Data-driven fault identification of ageing wind turbine based on NARX

As the existing wind turbine is approaching the designed service life, it is of great significance to check the ageing condition in advance. In this study, a system identification model, nonlinear autoregressive network with ex-ogenous inputs (NARX), was used to analyze the ageing condition of wind tur-bines. This data-driven approach uses the input and output data of the system directly without the need for specific mathematical models. Simulated experi-mental data for four different ageing conditions are used for system identifica-tion. By comparing the NARX model parameters under different conditions, the fault conditions of the system can be found and the degree of ageing can be de-tected.