Multi-fidelity Graph Neural Networks for Efficient Power Flow Analysis under High-Dimensional Demand and Renewable Generation Uncertainty

This paper presents a multi-fidelity graph neural network approach for probabilistic power flow analysis under high-dimensional demand and renewable-generation uncertainty. The work shows how graph-based surrogate models can accelerate repeated power-flow studies while remaining aligned with the structure and operating variability of modern power networks.

Recommended citation: Taghizadeh, M., Khayambashi, K., Hasnat, M. A., & Alemazkoor, N. (2024). "Multi-fidelity Graph Neural Networks for Efficient Power Flow Analysis under High-Dimensional Demand and Renewable Generation Uncertainty." Electric Power Systems Research.