Abstract: Temperature rise in the gearbox high-speed bearings is a key indicator of the operational status and health of 5 MW wind turbines. Given that high-speed bearing temperature rise is influenced by the coupled effects of multiple factors and exhibits strong nonlinearity, this paper proposes a regression prediction method for high-speed bearing temperature rise based on Bayesian-TCN and SHAP analysis. First, a time-series convolutional neural network （TCN） regression prediction model is constructed, and Bayesian optimization algorithms are employed to optimize its hyperparameters. Second, the SHAP method is introduced to interpret the importance of the model’s input features and their underlying mechanisms. Finally, simulation experiments are conducted on the MATLAB platform. The results demonstrate that this method achieves high prediction accuracy and can serve as a reference for wind turbine condition monitoring, anomaly early warning, and O&M decision-making.