Physics-Informed Graph Neural Network for Dynamic Reconfiguration of Power Systems

To maintain a reliable grid we need fast decision-making algorithms forcomplex problems like Dynamic Reconfiguration (DyR). DyR optimizes distributiongrid switch settings in real-time to minimize grid losses and dispatchesresources to supply loads with available generation. DyR is a mixed-integerproblem and can be computationally intractable to solve for large grids and atfast timescales. We propose GraPhyR, a Physics-Informed Graph Neural Network(GNNs) framework tailored for DyR. We incorporate essential operational andconnectivity constraints directly within the GNN framework and train itend-to-end. Our results show that GraPhyR is able to learn to optimize the DyRtask.