GridSFM: A Foundation Model for Solving AC Optimal Power Flow

2026-09-24

AI Summary

This study addresses the challenge in large-scale AC optimal power flow where continuous neural networks struggle to approximate solution mappings over disconnected feasible sets. We propose a framework integrating a pretrained foundation model with physics-informed fine-tuning. Employing a 15-million-parameter graph neural network, the method relaxes constraints via logarithmic penalties while elevating problem dimensionality. We theoretically establish the contractibility of elastic feasible sets, ensuring well-posedness when projecting approximate solutions back onto the original feasible set, and leverage Newton’s method for physics-informed fine-tuning. This approach overcomes generalization bottlenecks across complex topologies, achieving a mere 2.45% zero-shot error on ten-thousand-bus systems. Requiring only minimal data to adapt to unseen power grids, it outperforms dedicated single-topology models.

Abstract

We introduce GridSFM, a framework that combines a pretrained foundation model across grid topologies with physics-informed fine-tuning for solving AC Optimal Power Flow (AC-OPF) at scale. It is a $15$ million parameter physics-inspired graph neural network pretrained across $54$ topologies of $500$ to $4{,}000$ buses. Our model attains a $2.45\%$ zero-shot generation-cost error on a $10{,}000$ bus case held-out operating conditions with no degradation as system size grows. Building on this, we pair the pretrained backbone with a physics-informed fine-tuning design based on Newton's method for power flow. With only $100$ solved instances, GridSFM adapts to unseen grids up to $10{,}000$ buses. We show it out performs single topology, dedicated neural network models that are trained more data, both in terms of cost and solver iterations when deployed as warm starting points. In designing this foundation model, we overcome the fact that the feasible set for AC-OPF can be disconnected. This is an obstruction that prevents any continuous neural network from approximating the solution map. To do so, we lift the problem and relax its constraints with logarithmically penalized slacks. We prove that the resulting elastic feasible set is contractible, that the AC-OPF minimizers remain minimizers of the elastic problem above an explicit penalty threshold, and that projecting an approximate solution back onto the AC-OPF feasible set is well posed. We release all models, data, and code so that the community can build on a shared starting point for AC-OPF.

Problem

• AC Optimal Power Flow
• Foundation Model
• Grid Topology Generalization
• Disconnected Feasible Set
• Scalability

Innovation

• Foundation Model
• Physics-Informed Fine-Tuning
• Graph Neural Network
• AC Optimal Power Flow
• Elastic Feasible Set