Deep Learning for Sustainable Solar Energy Prediction

مجالات المشروع

وصف المشروع

The project develops a deep learning-based forecasting system for Global Horizontal Irradiance (GHI), which is an important indicator of solar energy availability. The system focuses on short-term solar irradiance forecasting for the Qassim region by integrating NASA POWER satellite data with ground-based meteorological measurements from K.A.CARE. The dataset covers the period from August 2021 to August 2022. The project applies data preprocessing, timestamp alignment, normalization, feature engineering, and 24-hour sliding-window sequence construction. Three deep learning models were developed and evaluated: LSTM, GRU, and CNN-LSTM. The models forecast GHI across three horizons: 1-hour, 3-hours, and 24-hours ahead. Model performance was evaluated using MAE, RMSE, and R². An interactive Streamlit dashboard was also developed to visualize live forecasts, compare model performance, and present dataset insights. The project supports sustainable solar energy planning and decision-making by providing accurate multi-horizon GHI forecasts.

فريق المشروع