Domain-specific ReAct for physics-integrated iterative modeling: A case study of LLM agents for gas path analysis of gas turbines
This addresses the challenge of integrating AI into energy and power engineering for gas turbine diagnostics, but it is incremental as it adapts existing ReAct/tool-calling methods to a specific domain.
This study applied large language models (LLMs) with tool-calling capabilities to gas path analysis of gas turbines, finding that larger models (like GPT) performed better than smaller ones (like LLama3) but all struggled with complex multi-component problems, suggesting that models with nearly 100 billion parameters could meet professional needs with fine-tuning.
This study explores the application of large language models (LLMs) with callable tools in energy and power engineering domain, focusing on gas path analysis of gas turbines. We developed a dual-agent tool-calling process to integrate expert knowledge, predefined tools, and LLM reasoning. We evaluated various LLMs, including LLama3, Qwen1.5 and GPT. Smaller models struggled with tool usage and parameter extraction, while larger models demonstrated favorable capabilities. All models faced challenges with complex, multi-component problems. Based on the test results, we infer that LLMs with nearly 100 billion parameters could meet professional scenario requirements with fine-tuning and advanced prompt design. Continued development are likely to enhance their accuracy and effectiveness, paving the way for more robust AI-driven solutions.