Scalable Heterogeneous Graph Foundation Models for Data-Driven Optimal Power Flow in Smart Grids
A new paper presents a scalable heterogeneous graph neural network workflow for optimal power flow in smart grids. This approach maintains the unique structures of power networks while enabling efficient training and optimization on supercomputers. The findings indicate improved accuracy and training stability through fine-tuning of pretrained models.
- ▪The paper introduces a scalable heterogeneous graph neural network workflow for data-driven optimal power flow surrogate modeling.
- ▪It utilizes three million heterogeneous graph instances across various grid topologies for training and optimization.
- ▪Fine-tuning pretrained models enhances accuracy and reduces adaptation costs in power flow applications.
Opening excerpt (first ~120 words) tap to expand
Computer Science > Machine Learning arXiv:2605.23194 (cs) [Submitted on 22 May 2026] Title:Scalable Heterogeneous Graph Foundation Models for Data-Driven Optimal Power Flow in Smart Grids Authors:Massimiliano Lupo Pasini, Yijiang Li, Kibaek Kim, Teja Kuruganti View a PDF of the paper titled Scalable Heterogeneous Graph Foundation Models for Data-Driven Optimal Power Flow in Smart Grids, by Massimiliano Lupo Pasini and 3 other authors View PDF HTML (experimental) Abstract:Fast and reliable optimal power flow (OPF) approximation is essential for reliable smart-grid operation, yet many learning-based surrogates either flatten the native heterogeneous structure of power networks, target a limited set of grid topologies, or lack scalable infrastructure for graph foundation model (GFM)…
Excerpt limited to ~120 words for fair-use compliance. The full article is at arXiv cs.AI.
Discussion
0 commentsMore from arXiv cs.AI
