A Novel Method for Renewable Energy Output Prediction Based on Transformer-Gated Recurrent Unit Model

Addressing the current challenges in the accuracy of renewable energy output prediction, this paper proposes a novel method for renewable energy output prediction based on a Transformer-Gated Recurrent Unit(Transformer-GRU)model. Initially, predictions of renewable energy output are conducted separately using Transformer and GRU models to evaluate their predictive performance. Subsequently, a combined Transformer-GRU model is employed for prediction. Finally, the predictive performances of the various models are compared. The results demonstrate that the renewable energy output prediction method based on the combined Transformer-GRU model exhibits greater accuracy and superiority. With the MAE value and the $\mathbf{R}^{\mathbf{2}}$ value being 0.9744 and 0.9183 respectively, The prediction effect of the Transformer-GRU model has increased by $13.6 \%$ and $41.7 \%$ respectively compared with the individual prediction effects of the Transformer model and the GRU model.