Solar Power Prediction

Developed a machine learning-based solar power prediction system that forecasts photovoltaic (PV) output using meteorological and historical irradiance data, enabling better grid integration and energy planning.

Implemented and compared multiple ML models including LSTM-based recurrent neural networks, Random Forest, and XGBoost to predict hourly and day-ahead solar generation, evaluating models based on RMSE and MAE metrics.

The LSTM-based approach achieved the best prediction accuracy, capturing temporal dependencies in solar irradiance patterns across varying weather conditions and seasons, making it suitable for practical deployment in smart grid environments.