StorytellingSerious

NeurIPS · AI Foundations for Power Grids Workshop · First posted

FLARE: Physics-Grounded Evaluation of Language-Model Agents for Feeder Restoration with Degraded Observations

BHVSP Subrahmanyam, Neeraj Choudhary

Accepted · Workshop · Agentic AI

FLARE tests how language-model agents restore electrical feeders with incomplete observations. In simulation, shadow replay found that 96.4% of 15,706 guard-rejected actions would have violated a physical constraint.

Abstract

Restoring service after distribution-grid faults requires decisions from incomplete and conflicting information while respecting electrical constraints. Addressing this need, we introduce FLARE (Feeder-Level Agent Restoration Evaluation), a physics-grounded benchmark for post-isolation feeder restoration. FLARE contains 800 AC-feasible cases across five observation settings and allows agents to gather evidence, evaluate switches, and perform restoration actions under fixed operational limits. We evaluate five instruction-tuned models over 78,750 episodes, including topology, distributed-generation, and loading shifts. Mean normalized regret ranges from 0.082 to 0.148. Although every model has zero median regret, P95 regret ranges from 0.814 to 1.000, revealing an important failure tail. Shadow replay shows that 96.4% of 15,706 guard-rejected actions would have caused a physical violation. FLARE enables reproducible evaluation of language-model-based restoration support while emphasizing the need for external safety checks. Code is available at https://github.com/dr-pandit-69/FLARE-neurips.

Findings

  • Evaluates five instruction-tuned models across 78,750 episodes and 800 AC-feasible cases.
  • In simulation, shadow replay found a physical violation for 15,142 of the 15,706 guard-rejected actions (96.4%).
  • All five models had zero median regret, yet their 95th-percentile regret ranged from 0.814 to 1.000.

What the evaluation covers

The simulator models balanced power flow. It does not test unbalanced three-phase faults or short-circuit behavior. The result describes actions replayed in simulation. Deployment would need separate operator and equipment validation.

Read the methods and limitations in the paper (v1).

My contribution

Led the research and experimental development of this project.

Sources and citation

View BibTeX citation
@misc{flare-2026,
  title = {{FLARE: Physics-Grounded Evaluation of Language-Model Agents for Feeder Restoration with Degraded Observations}},
  author = {BHVSP Subrahmanyam and Neeraj Choudhary},
  year = {2026},
  eprint = {2610.23251},
  archivePrefix = {arXiv},
  primaryClass = {cs.LG},
  url = {https://arxiv.org/abs/2610.23251},
  note = {Accepted at NeurIPS · AI Foundations for Power Grids Workshop}
}

What are you in the mood for?