Large Models for Machine Monitoring and Fault Diagnostics:  Opportunities, Challenges, and Future Direction

Xuefeng Chen; Yaguo Lei; Yanfu Li; Simon Parkinson; Xiang Li; Jinxin Liu; Fan Lu; Huan Wang; Zisheng Wang; Bin Yang; Shilong Ye; Zhibin Zhao

doi:10.37965/jdmd.2025.832

Authors

Xuefeng Chen School of Mechanical Engineering, Xi’an Jiaotong University, Xi’an, China
Yaguo Lei School of Mechanical Engineering, Xi’an Jiaotong University, Xi’an, China
Yanfu Li Department of Industrial Engineering, Tsinghua University, Beijing, China
Simon Parkinson Department of Computer Science, University of Huddersfield, Huddersfield, UK
Xiang Li School of Mechanical Engineering, Xi’an Jiaotong University, Xi’an, China https://orcid.org/0000-0003-0569-2176
Jinxin Liu School of Mechanical Engineering, Xi’an Jiaotong University, Xi’an, China
Fan Lu School of Mechanical Engineering, Xi’an Jiaotong University, Xi’an, China
Huan Wang Department of Industrial Engineering, Tsinghua University, Beijing, China https://orcid.org/0000-0002-1403-5314
Zisheng Wang Department of Industrial Engineering, Tsinghua University, Beijing, China
Bin Yang School of Mechanical Engineering, Xi’an Jiaotong University, Xi’an, China https://orcid.org/0000-0002-3015-3580
Shilong Ye Department of Industrial Engineering, Tsinghua University, Beijing, China
Zhibin Zhao School of Mechanical Engineering, Xi’an Jiaotong University, Xi’an, China https://orcid.org/0000-0003-4180-7137

DOI:

https://doi.org/10.37965/jdmd.2025.832

Keywords:

context learning; fault diagnostics; LLMs; multimodal learning

Abstract

As a critical technology for industrial system reliability and safety, machine monitoring and fault diagnostics have advanced transformatively with large language models (LLMs). This paper reviews LLM-based monitoring and diagnostics methodologies, categorizing them into in-context learning, fine-tuning, retrieval augmented generation, multimodal learning, and time series approaches, analyzing advances in diagnostics and decision support. It identifies bottlenecks like limited industrial data and edge deployment issues, proposing a three-stage roadmap to highlight LLMs’ potential in shaping adaptive, interpretable PHM frameworks.