Graph Neural Networks on Factor Graphs for Robust, Fast, and Scalable Linear State Estimation with PMUs

• DOI: 10.48550/arxiv.2304.14680
• Publication Date: 2023-01-01
• Authors: Ognjen Kundačina; Mirsad Ćosović; Dragiša Mišković; Dejan Vukobratović
• Keywords: Unobservable; Robustness; Units of measurement; Factor graph; Phasor measurement unit

As phasor measurement units (PMUs) become more widely used in transmission power systems, a fast state estimation (SE) algorithm that can take advantage of their high sample rates is needed. To accomplish this, we present method uses graph neural networks (GNNs) to learn complex bus voltage estimates from PMU and current measurements. We propose an original implementation GNNs over the system's factor simplify integration various types quantities measurements on system buses branches. Furthermore, augment improve robustness GNN predictions. This model highly efficient scalable, as its computational complexity linear with respect number nodes system. Training test examples were generated by randomly sampling sets annotated exact solutions SE PMUs. The numerical results demonstrate provides accurate approximation solutions. errors caused malfunctions or communication failures would normally make problem unobservable have local effect do not deteriorate rest