Abstract for: Multi-Timescale Power System Dynamics: Leveraging Large Language Models for Enhanced System Dynamics Modeling

This paper demonstrates a novel methodological approach that leverages Large Language Models (LLMs) to enhance the development of system dynamics models for complex multi-timescale systems. Through a case study of electrical power systems, we utilized Claude, an advanced AI assistant, to construct, refine, and document three conceptual models representing electromagnetic (microseconds), electromechanical (seconds), and operational (hours) timescales. While these models are necessarily simplified representations designed for educational and methodological demonstration purposes, they effectively illustrate the distinct dynamics across timescales and help identify potential cross-scale interactions. Our results show that AI-assisted modeling significantly improves efficiency in model formulation (reducing development time by approximately 60%), facilitates rapid debugging of structural errors, and enhances knowledge integration across disciplines. The primary limitations include dependence on human verification for physical validity and limited capability for novel conceptual innovation. The approach demonstrates particular value for educational contexts and as a foundation for more technically detailed implementations. As described in paper, Perplexity Deep Research and Claude Sonnet for model development assistance