Abstract:The main shaft thrust bearing is a crucial component of the wind turbine, and any failure can lead to severe losses for the entire unit. To achieve early fault warning for the main shaft thrust bearing of wind turbine generator sets and take timely maintenance measures to avoid further escalation of faults, this paper focuses on the temperature of the wind turbine''s main shaft thrust bearing. It proposes a Stacking-based fault early warning model using data acquisition and supervisory control (SCADA) under normal operating conditions of wind turbine generators. Firstly, this study utilizes the ratio of goodness of fit and mean square error of four individual models to comprehensively rank the features, resulting in four sets of datasets with different combinations of gradient features. Secondly, by analyzing the predictive performance and correlation of the single model, XGBoost, LightGBM, and Random Forest are selected as the base learners, with XGBoost as the meta-learner to establish the Stacking ensemble learning prediction model. Experimental results show that the Stacking model based on temperature prediction of the main shaft thrust bearing performs the best, with significantly improved prediction errors compared to the base learners. Finally, the root mean square error (RMSE) of the temperature prediction model is calculated, and the error threshold for the normal state of the main shaft thrust bearing is set based on the Exponential Weighted Moving Average (EWMA) method. Experimental results demonstrate that the Stacking model established in this paper can issue fault warnings for wind turbines'' main shaft thrust bearing at least 6 hours in advance.