SELECTION OF VARIABLES IN DIGITAL TWIN VALIDATION: EVIDENCE, APPROACHES, AND METHODOLOGICAL GAPS

The increasing adoption of Digital Twins in industrial settings has intensified the need for effective validation and continuous monitoring strategies. In this context, the definition of the variables used to assess the adherence between the model and the real system becomes a critical factor, still little explored in a structured way in the literature. This paper presents an excerpt from a systematic review of the broader literature, developed within the scope of a doctoral thesis, with the objective of analyzing how the selection of variables has been approached in the context of validation and monitoring of digital twins. The review was conducted according to the PRISMA protocol, resulting in the analysis of 51 studies. Different approaches to variable selection were identified, including expert-based methods, statistical techniques, Machine Learning, qualitative approaches, and ad hoc strategies. The results show that, despite the diversity of methods, the selection of variables occurs predominantly implicitly, without consolidated guidelines.