State Estimation in Electric Power Systems Leveraging Graph Neural Networks

• DOI: 10.36227/techrxiv.18131207.v2
• Publication Date: 2022-04-11T14:35:55Z
• Authors: Ognjen Kundacina; Mirsad Cosovic; Dejan Vukobratovic
• Keywords: Solver; Units of measurement; Phasor measurement unit

The goal of the state estimation (SE) algorithm is to estimate complex bus voltages as variables based on available set measurements in power system. Because phasor measurement units (PMUs) are increasingly being used transmission systems, there a need for fast SE solver that can take advantage high sampling rates PMUs. This paper proposes training graph neural network (GNN) learn estimates given PMU voltage and current inputs, with intent obtaining accurate predictions during evaluation phase. GNN trained using synthetic datasets, created by randomly sets system labelling them solution obtained linear PMUs solver. presented results display accuracy various test scenarios tackle sensitivity missing input data.