Wissensmanagement für Wartung und Instandhaltung im Verteilnetz – Konzeption eines Assistenzsystems basierend auf einem Large Language Model

Distribution grid operators in Germany are facing major challenges in managing organization-specific knowledge: staff shortages due to demographic change, knowledge is only implicitly available and not digitalized in knowledge management systems, in some cases there are no knowledge management systems or concepts at all, and the distribution grid is becoming increasingly complex. In conjunction with the increasing stress on central components in the distribution grid due to the energy transition, new solutions are needed, especially for knowledge-intensive maintenance and repair processes. Generative artificial intelligence as an emerging technology, in particular through large language models, is showing initial success for guidance, decision support, and knowledge transfer. Building on the design science research paradigm, this article develops a holistic approach to knowledge management based on an assistance system as a central component. A large language model is used to assist technicians on the distribution grid during maintenance and repair activities based on manuals. In addition to a conceptual design, this article shows the strategy developed for demonstrating and evaluating the results. The article provides a novel concept for distribution grid operators through assistance systems for knowledge management based on large language models and shows downstream steps that are necessary before a market launch.