Renewable Energy Consumption on Solar and Wind Energy Prediction Using Deep Learning and Generative AI

Solar and wind energy harvest electricity without producing harmful greenhouse gases or pollutants, considerably dipping their environmental impact when measured against fossil fuels. To forecast solar and wind energy production this paper suggests a method that chooses attributes with a strong contribution on the impact of prediction of energy production by ARIMA, Ridge and Lasso Regression. Time series data was obtained from open power systems for 37 European countries. Most importantly, improving the performance of the proposed model without regard to deep learning models is explored. In order to get insights into how the predictions are made by LSTM, the Explainable AI (XAI) method is used. The obtained local approximations to the model's decision boundary on certain predictions for specific instances. In addition, data from decision trees, random forests, and support vector machines (SVM) are used to validate it. 100% of the data was taken into consideration for validation after being split into 0.8% and 0.2 %, respectively, for training and testing. As the prediction values the amount of energy consumption of the year 2025 in solar is 72530 MW and wind energy is 45280 MW.