GridFM
Small Foundation Models for the Electric Grid
GridFM brings a new class of modeling capabilities to power‑system planning and operations by combining physics, data, and foundation‑model architectures. Instead of relying on slow, case‑by‑case numerical solvers, GridFM provides a fast, generalizable representation of the electric grid that can be adapted to different networks, operating conditions, and study objectives
1. A Foundation Model Trained on Grid Physics
At its core, GridFM learns the structure and behavior of power networks from large collections of grid states and physics‑based simulations. This enables the model to deliver accurate, physically consistent inferences in milliseconds, supporting workflows that previously required minutes or hours of computation.
The model learns fundamental relationships among:
- Voltage magnitudes
- Phase angles
- Line flows & thermal limits
- Generator dispatch
- Load variations
- Topological or cost perturbations
This FM becomes a general-purpose, reusable base model for many downstream tasks.
2. Fine‑Tuning for Specialized Applications
Because the model is trained as a general foundation model, GridFM can be rapidly fine‑tuned to reflect local topologies, operational practices, or proprietary datasets. Organizations can adapt the model to their proprietary networks using limited labeled data, enabling:
- Custom OPF solvers
- Islanding analysis
- Dispatch optimization
- Planning studies
- Emergency operations support
- Real-time reliability evaluation
3. Fast & Accurate Inference
These capabilities make GridFM a unified modeling layer for the modern electric grid. GridFM targets orders‑of‑magnitude faster performance than classical solvers, enabling:
Online optimization under uncertainty
Real-time contingency ranking
Rapid n‑1/n‑k analysis
Multi-scenario planning