Project Grant 2334448

This $225,000 National Science Foundation (NSF) Project Grant award under the CFDA 47.041 Engineering program aims to develop a physics-informed, real-time optimal power flow model using machine learning techniques. The project seeks to address gaps in providing close to optimal solutions for power plant outputs while considering practical dynamical constraints to avoid frequency fluctuations and grid instabilities. The key scientific and engineering contributions include: (1) advancements in combining physics-informed neural networks with conventional feed-forward neural networks to predict solutions to optimal power flow problems in real-time; (2) novel approaches to ensure constraint satisfaction in the learned embedding; (3) techniques to ensure scalability of training to large, realistically-sized networks; and (4) pursuit of model robustness by assessing performance under measurement noise and analyzing reliability. This project aims to expedite the adoption of increased renewable energy into the power grid by reducing curtailment resulting from stability concerns and suboptimalities from conventional heuristic droop control. The award was made to the University of Virginia, an educational institution that has extensive experience delivering federal research projects across fields like engineering, physics, and computer science.