Automated Adaptation of Digital Twins for Production Line Design by Adapting Existing Models Through LLM-Guided Expert Interviews
Year: 2025
Type: conference-paper
Abstract: The design of new production lines is a costly process, considering multiple variants must be compared.Simulations are indispensable to evaluate different production concepts.However, they require detailed models.A digital twin, which can be used not only in the design phase but also in operation, offers potential here.The creation of such a twin is often complex, as it requires a complex modelling phase.This paper presents a framework to automate the creation of digital twins for production line design by adapting existing models through Large Language Model-guided expert interactions.New production lines often are similar to existing ones, for which digital twins can be generated by process mining.The adaptation of these models is then carried out by a domain expert using a Large Language Model that allows to make changes to the digital twin without greater modelling knowledge.The framework is evaluated using a case study from the "IoT-Factory" at our university.The results show that it is suitable for reducing the effort needed to create digital twins for production line design by adapting existing models.
DOI: 10.25368/2025.247
Authors: Lennart Schipper, Christian Schwede
Institution: Hochschule Bielefeld
Language: English
Subfield: Industrial and Manufacturing Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W4415322973