AI Boosted Transformer Life Management

Models (LLMs) in 2022, it became clear that Artificial Intelligence (AI) Engines can be successfully trained to “learn” from an almost limitless amount of technical and non-technical material, on any subject matter, of public or private domain or a combination of both. Given the broad availability of such LLMs, the solution described in this paper is therefore based on four fundamental pillars : first, the technical knowledge base; that is a collection of papers, guides and publications in general, that together with other internationally recognized organization, such as IEEE, CIRED, and others, provide the AI engine with the necessary foundational knowledge, including statistics, to learn from; second comes a database, as a collection of transformers operational data with the users’ actual experience of life-long healthy and unhealthy units; the next pilar and at the core of the solution is the AI Engine that learns from the knowledge base and permanently adapts and balances its own knowledge to the dynamic nature of the users’ database. Finally, a query system through which the AI Engine will then interact with the end-user, responding to fundamental questions about their actual transformers condition, from design aspects to manufacturing, from maintenance to operations, from normal to abnormal behavior based on historical data, the knowledge base and stats from the database using also the knowledge extracted from similar units as a means of robust statistical comparative analysis.

The paper provides a real-life demonstration of such engine, trained with appropriate technical material about transformers operating conditions and additional technical references that may support transformer condition assessment.