Deep Learning-Based Prediction of Solar and Wind Energy for Smart Grid Optimization

With the fast adoption of renewable energy sources such as solar and wind, ensuring power system reliability requires accurate short-term forecasting of generation output. This paper investigates the application of AI-based modelsincluding ensemble and deep learning techniques for predicting renewable energy generation. The models examined comprises of Linear Regression, Random Forest, LSTM, GRU, and a hybrid CNN-LSTM framework. Using historical meteorological and power generation data, we evaluated each model's accuracy with parameters like RMSE, MAE, and R2. The results show that hybrid deep learning models, i.e. CNN-LSTM, performs better than traditional methods. They effectively model nonlinear and time-related patterns in weather-influenced data. These findings show the promise of hybrid AI frameworks in supporting dependable and efficient renewable energy integration for smart grid systems.