An Experimental Study on Renewable Energy Forecasting and Prediction Using Deep Learning

The study compares different RNN, LSTM, and VGG16 models to predict renewable energy production. Thus, according to their results, the LSTM and VGG16 are models with the highest abilities to deal with solar and wind energy, which is intermittent, especially using the year 2020 data. In particular, from these models, LSTM and VGG16 models were found to provide outstanding capabilities for the quite fluctuating output of renewable energy. Therefore, LSTM and VGG16 appear to show good robustness in dealing with variability in renewable energy output and tend to be more practical for real scenarios requiring accurate forecasting. Besides, it compared past papers with this one, which did not reach the level of accuracy of this paper. In contrast, our implementation was remarkably successful in comparison, especially with the best score achieved for LSTM. Comparing efficiency with RNN, reaching only a moderate accuracy of 74%, means it proves superiority and natural ability in using models like LSTM and VGG16, especially for grid management and the prediction strategies used in energy.