Physics-informed machine learning for identification and scalable control of integrated multi-energy systems
The integration of electricity, gas, and heating/cooling networks into multi-energy systems (MESs) is increasingly recognized as a key step towards efficiency, flexibility, and decarbonization. However, modeling and controlling such systems remains challenging due to their large scale, nonlinear dynamics evolving at different time scales, and limited sensing infrastructures. Traditional physics-based models are often too complex, while purely data-driven methods require large datasets and may disregard fundamental physical laws.This research proposes a systematic approach that combines physical knowledge with data to build consistent, compact, and generalizable models of MESs, even in the presence of partial measurements. At the same time, scalable control strategies are designed to ensure stability and optimize performance across interconnected subsystems. The methodology will be validated through both simulated scenarios and laboratory benchmarks, with the goal of delivering innovative tools for the reliable and sustainable management of future energy networks.
Back to Current Students