Abstract: In order to identify the abnormal running state of the generator in time, a wind turbine generator front bearing fault early warning method based on Bayesian optimized extreme gradient boosting algorithm is proposed. The historical data collected by SCADA (Supervisory Control And Data Acquisition) are preprocessed by effective data preprocessing methods. The temperature prediction model of the front bearing of wind turbine generator is constructed based on the Bayesian-optimized XGBoost (eXtreme Gradient Boosting) algorithm and the fault early warning threshold of the front bearing of the wind turbine generator is determined based on the 3σ criterion. The experimental results show that the proposed method can detect the abnormal signals of the front bearing of the wind turbine generator in advance. Compared with the models established by random search and grid search, the advantages of Bayesian optimization model in generalization performance and prediction accuracy are verified.

Key words: XGBoost (eXtreme Gradient Boosting) algorithm, wind turbine, failure early warning, Bayesian optimization