Digital Twins in manufacturing: A Systematic Literature Review with Retrieval-Augmented Generation

This paper presents a systematic literature review on the use of digital twins in manufacturing, with the goal of developing a comprehensive taxonomy that synthesizes existing categorizations. Given the increasing complexity and volume of literature in this domain, conventional review methods are becoming insufficient. To address this challenge, the study applies a novel approach named retrieval augmented generation. This is a technique that combines large language models with real-time information retrieval, enabling the automated identification and summarization of typologies across a broad corpus of publications. A total of 1,354 publications were initially screened, leading to 144 distinct categorizations relevant to digital twins in industrial contexts. The resulting taxonomy classifies digital twins along multiple dimensions, including life cycle stages, physical domain and hierarchy levels, model characteristics, digital thread connectivity and deployment strategies. This work provides both researchers and practitioners with a structured approach to understanding and implementing digital twins in manufacturing environments, as well as a guideline to completely describe a specific implementation. The taxonomy serves as a foundation for future research and as a practical tool for industrial applications, since it defines design decisions, which have to be made.