Prediction Method of Eddy Current Losses in Shield Sleeve Failure Based on Bagging Ensemble Learning

Aiming at the insufficient research on eddy current loss prediction methods after the failure of the main pump motor shield sleeve, this paper proposes a prediction method based on Bagging ensemble learning. By integrating seven heterogeneous base learners, including decision tree regression, support vector machine regression, and random forest regression, an ensemble model is constructed to predict eddy current losses using features such as voltage, current, torque, and rotational speed. Experiments are conducted on a dataset generated by finite element simulation, covering shield sleeve failure types (wear, rupture, bulging) and varying degrees of failure. The prediction performance under low-speed and high-speed operating conditions is compared. The results show that the Bagging ensemble model outperforms single models in terms of mean squared error (MSE), root mean squared error (RMSE), and coefficient of determination (R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup>). The MSE reaches as low as 0.0043 (low-speed condition) and 0.0069 (high-speed condition), with R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> peaking at 0.9949, validating the high accuracy and robustness of the method in complex failure scenarios. This study provides an effective tool for shield sleeve fault monitoring and loss optimization.