What is it about?
This study introduces a new way to manage home solar panels and batteries to make energy use more efficient and support the power grid. By combining math-based models with advanced machine learning, we predict how much energy a home will produce and use. We also developed two smart strategies to control when batteries charge or release energy, helping homes use more of their solar power and reduce strain on the grid. Testing different setups showed that the right balance of solar panels and battery size is key to maximizing benefits, especially when considering local weather and energy needs.
Featured Image
Photo by Zaptec on Unsplash
Why is it important?
Our research introduces a unique hybrid approach that blends physics-based models with advanced machine learning (XGBoost) to optimize home solar-battery systems, a timely solution as renewable energy adoption grows globally. Unlike traditional methods that prioritize cost over grid stability, our framework balances both, using innovative control strategies (PPO and RBC) to enhance energy self-use and reduce grid fluctuations. By testing across varied solar and battery sizes, we provide new insights into optimal system configurations, addressing a critical gap in scalable, grid-friendly renewable energy solutions. This work could drive more efficient home energy systems, supporting the global push for sustainable energy and stable power grids.
Perspectives
Guiding my PhD student, Xin Liu, through this publication was a deeply fulfilling journey, blending our shared passion for sustainable design with innovative building operations. As architects, we approached this research from a unique perspective, focusing on net-zero energy buildings through optimized solar-battery systems—an area typically dominated by electrical engineering experts. This cross-disciplinary approach enabled us to rethink energy management in homes, prioritizing both environmental impact and grid stability. I’m excited by how this work challenges conventions in our field and hope it inspires architects to lead in designing net-zero, grid-friendly buildings that shape a sustainable future.
Zhonghua Gou
Wuhan University
Read the Original
This page is a summary of: Hybrid forecasting and optimization framework for residential photovoltaic-battery systems: Integrating data-driven prediction with multi-strategy scenario analysis, Building Simulation, July 2025, Tsinghua University Press,
DOI: 10.1007/s12273-025-1319-6.
You can read the full text:
Resources
Contributors
The following have contributed to this page
