Large language models safely support electric grid operators decisions
A Governance-Aware Large Language Model Orchestrated Agentic Digital Twin for Transmission System Operator Control Room Decision Support
Software EngineeringArtificial Intelligence
Summary
Electric grid operators face new challenges as renewable energy grows and the system becomes more complex. The authors developed a digital assistant powered by a large language model that can choose safe, approved analysis tools to help operators without making mistakes. A governance system ensures the assistant only runs allowed tools, sticks to limits, requires human approval for risky actions, and only reports numbers verified by real data. This helps operators trust the assistant’s advice while keeping control and safety intact.
What this means in practice
- •For power grid operators: Use a controlled AI assistant that enforces safety rules to support complex grid management decisions with trustworthy data and operator approval.
- •For industrial control system engineers: Integrate governance-aware AI models in control rooms to automate analysis choosing only verified tools and maintaining traceability for critical infrastructure.
Authors
Costas Mylonas, Magda Foti, Emmanouel Varvarigos
Abstract
Transmission system operators face rising complexity from renewable integration, reduced inertia, and tighter security margins. Large language models offer natural-language decision support, but their hallucinations, uncontrolled tool use, and weak traceability conflict with control room requirements. This paper presents a governance-aware agentic digital twin for transmission grid control rooms. The large language model only selects and parameterizes whitelisted analysis tools, and every proposed action passes through a governance layer that the model cannot bypass. The layer enforces four rules on every run. Only whitelisted tools execute. No run exceeds its step budget. No action with side effects executes without explicit operator approval. Every number in an answer is rendered by the layer from backend results with its unit, variable, and time. The rules are checked on a persistent audit trail for every run of a released 118-task benchmark, which covers analytics, simulation, multi-step workflows, and twelve families of adversarial inputs on a digital twin of the Greek transmission network. Across 590 runs of the primary model, tool selection reaches 96.5% and task success 93.7%, and all four rules hold without exception. In a separate three-repetition study across four large language models, 1416 runs in total, the rules again hold on every run, with an approximate 95% lower bound of 99.8%. Removing the layer makes the same model execute all 45 approval-requiring runs without authorization and leaves only 39.2% of its answers with backend-supported numbers. Enforcement costs 12 to 16 milliseconds per request.