Power System Simulation Intelligent Assistant: A Chatbot Integrating Large Language Models and Domain Models

Simulations are crucial for secure and economical operation of power systems. Recently, with the rapid advancement of artificial intelligence technologies, large language models (LLMs) like GPT-4 have demonstrated impressive capabilities in natural language processing tasks and their applications are being widely explored in many fields. However, directly applying LLMs in power system simulations is challenging due to their limitations in specialized domain knowledge. Despite this, LLMs offer significant flexibility in interactions, making them promising candidates for collaboration with existing power system domain models (PSDMs). To facilitate this integration, a collaboration framework between LLMs with PSDMs is proposed and a power system simulation intelligent assistant is developed based on this framework. The assistant guides users through power system simulations in a logical sequence with an interaction-driven exception-handling mechanism to improve robustness. Furthermore, the assistants based on different LLMs are compared.