Applications of Large Language Models in Microgrids: A Literature Review

This paper reviews the growing body of studies exploring the application of Large Language Models (LLMs) in microgrids and smart grids and explains why practicing such models matters in the domain. The reviewed works are organized into five main categories: load forecasting, energy management systems, cybersecurity, fault detection, and agentic frameworks. Within these areas, LLMs have been applied both directly, for control, dispatch, and diagnostic tasks, and indirectly, as reasoning or optimization assistants. Although the research in this domain is still in its early stage, findings demonstrate strong potential for LLMs to enhance decision-making, automation, and user interaction across grid operations. The paper also discusses key challenges and directions for future work, including real-time deployment, model adaptation, and the development of multi-agent strategies for grid control and management.