Inroduction

This study investigated the effectiveness of three machine learning time series models—GRU, LSTM, and Transformer—for forecasting solar energy consumption. The models were trained and evaluated on a solar energy consumption dataset, and their performance was compared using metrics such as MSE, RMSE, MAE, and R2-score. The results demonstrated that The Transformer model achieved the highest accuracy and lowest error metrics, indicating its superior ability to capture long-range dependencies in solar energy consumption While not as accurate as the Transformer, GRU, and LSTM still provided reasonably good forecasting performance, making them suitable alternatives when computational resources are limited.