From Natural Language to Solver-Ready Power System Optimization: An LLM-Assisted, Validation-in-the-Loop Framework

2025-08-11

AI Summary

Large language models (LLMs) often generate infeasible or suboptimal solutions when directly applied to power system optimization, due to semantic ambiguities and lack of formal constraint adherence. Method: We propose a “natural language → mathematical model → reliable solution” verification-closed-loop framework. It integrates domain-specific prompt engineering, formal constraint validation, and a solver-guided iterative repair algorithm to enable LLMs to accurately generate executable mathematical programming models—particularly mixed-integer linear programs (MILPs)—and automatically verify feasibility and optimality. Contribution/Results: Unlike prior work treating LLMs as solvers, we reposition them as “modeling assistants,” leveraging human-in-the-loop collaboration to ensure semantically faithful translation from natural language to formal mathematics. Evaluated on unit commitment, our method consistently produces optimal or near-optimal schedules with 100% solution feasibility and cost deviation under 0.5%, substantially enhancing the reliability and practical applicability of automated modeling in power systems.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: (Large) Language Models
• Planning, Routing, and Scheduling: Planning with Language Models

Application Category

• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

This paper introduces a novel Large Language Models (LLMs)-assisted agent that automatically converts natural-language descriptions of power system optimization scenarios into compact, solver-ready formulations and generates corresponding solutions. In contrast to approaches that rely solely on LLM to produce solutions directly, the proposed method focuses on discovering a mathematically compatible formulation that can be efficiently solved by off-the-shelf optimization solvers. Directly using LLMs to produce solutions often leads to infeasible or suboptimal results, as these models lack the numerical precision and constraint-handling capabilities of established optimization solvers. The pipeline integrates a domain-aware prompt and schema with an LLM, enforces feasibility through systematic validation and iterative repair, and returns both solver-ready models and user-facing results. Using the unit commitment problem as a representative case study, the agent produces optimal or near-optimal schedules along with the associated objective costs. Results demonstrate that coupling the solver with task-specific validation significantly enhances solution reliability. This work shows that combining AI with established optimization frameworks bridges high-level problem descriptions and executable mathematical models, enabling more efficient decision-making in energy systems

Problem

• Convert natural-language power system descriptions to solver-ready formulations
• Ensure mathematically compatible formulations for optimization solvers
• Enhance solution reliability with AI and validation-in-the-loop

Innovation

• LLM-assisted natural language to solver-ready formulation
• Validation-in-the-loop ensures feasibility and optimality
• Domain-aware prompts integrate AI with optimization solvers