Improved Machine Learning Hybrid Approach For Advanced Solar PV Power Forecasting

Jessy Matar , Ahmed Elsherif und Markus Duchon

IEEE SusTech 2025 Conference,

April 2025 · Los Angeles, CA, USA · DOI: 10.1109/SusTech63138.2025.11025665

Zusammenfassung

Accurate forecasting of solar photovoltaic (PV) power generation is essential to optimize grid operations, improve energy management, and ensure the stability of renewable energy systems. In this study, we present a new improved hybrid machine learning approach to advance PV power forecasting. The proposed method integrates multiple machine learning models, including ensemble techniques and deep learning, to capture both short-term variability and long-term trends in solar power generation, making use of weather predictions to select the optimal forecasting model. By incorporating weather data, historical power outputs, and advanced feature selection, the model significantly improves the accuracy of power prediction. Through extensive testing on real-world photovoltaic datasets, our approach outperforms conventional forecasting techniques in terms of precision and reliability. The improved hybrid model’s enhanced accuracy offers valuable insights for grid operators and energy planners, contributing to more efficient integration of solar power into the energy grid.

Stichworte: PV, Forecasting, Machine Learning, Power, Accuracy

Url: https://ieeexplore.ieee.org/document/11025665