Graph neural networks, power flow problem, synthetic data, solution existence region
Abstract
This paper considers the application of graph neural networks for solving the problem of calculating steady-state modes of electrical power systems. A mathematical formulation based on representing the power system as a graph is proposed, and a model architecture combining local regression of node parameters and global classification of mode stability is developed. Synthetic data generated based on a standard test scheme were used for training. The conducted computational experiments demonstrated high accuracy in approximating the mode parameters. The obtained results confirm the promising potential of applying graph neural networks for accelerated analysis of electrical power system modes.
Author Biographies
Vladimir Anatol'evich Surin, South Ural State University, Chelyabinsk
Cand. Sc. (Engineering), Associate Professor at the Center of Excellence in AI, “VirtUm” – a Top-Tier Educational Program
Sergey Eduardovich Ivanov, South Ural State University, Chelyabinsk
Undergraduate Student, Department of Applied Mathematics and Programming
Maksim Eduardovich Ivanov, South Ural State University, Chelyabinsk
Undergraduate Student, Department of Applied Mathematics and Programming
Maksim Eduardovich Ivanov, South Ural State University, Chelyabinsk
Undergraduate Student, Department of Applied Mathematics and Programming
Valeriy Ivanovich Safonov, South Ural State University, Chelyabinsk
Cand. Sc. (Physics and Mathematics), Associate Professor of the Department Electric Power Generation Stations
Sergey Pavlovich Kulik, South Ural State University, Chelyabinsk
Dr. Sc. (Physics and Mathematics), Principal Researcher of the Research & Innovation Services
Nikolay Vladimirovich Maletin, South Ural State University, Chelyabinsk
