Generative AI in E-maintenance: Myth or Reality?

With increasing requirements for reliability, availability, efficiency, effectiveness, productivity, and security of the system, the importance of diagnostics and maintenance is also increasing. E-maintenance as a leading concept for maintenance management has so far primarily involved the use of domain-specific technical language processing (TLP) techniques on historical case data. Due to its popularity, generative AI (GAI) with large language models (LLMs) is starting to be used more and more in various technical areas, thus starting to take an increasingly important place in diagnostics and maintenance. Starting from the fact that the rapid development of information and communication technologies (ICT) was the main factor in the emergence and development of the concept of e-maintenance, the importance of the potential more serious application of all forms of generative AI in the context is clear. This is especially pronounced in cases of difficult or impossible access to the location of components or an uncertain situation related to the type of process (e.g., nuclear, aeronautical, space, offshore). Autonomous vehicles, vessels, and aircraft (as an indispensable part of today’s intelligent transport systems) are certainly a leading example of these cases. Regardless of the level of autonomy, these systems are extremely complex and difficult to maintain and represent a clear challenge for the application of new approaches. Therefore, the authors of the paper propose the use of middleware that would enable the integration of various GAI tools, algorithms, and models to increase the effectiveness of diagnostics and maintenance as close as possible to real-time. However, the exact extent of the possibilities and limitations of this approach has yet to be determined.