Vers une maîtrise des incertitudes des Jumeaux Numériques des systèmes industriels

Digital Twins (DTs) have emerged as essential tools, offering advanced capabilities for modeling, simulation, and optimization of industrial systems. They ensure seamless synchronization with the corresponding Physical System (PS), thereby supporting decision-making through analysis, diagnostics, and predictive functionalities. However, mastering uncertainty crucial for enhancing the robustness and resilience of critical industrial systems remains a major challenge in the development and operation of DTs. Indeed, while uncertainty management is vital, its full mastery has been rarely addressed in the literature. These uncertainties arise from various sources, including sensors, data, models, communication means, and operational conditions. Their impact varies depending on the DT’s maturity level and the phase of its lifecycle. Poor management of these uncertainties can compromise the fidelity of the DT with respect to its PS. In this work, we propose a structured mapping of uncertainties in the context of DTs along three complementary axes. First, we establish the correspondence between uncertainty types and a five-dimensional conceptual model of the DT. Second, we illustrate how uncertainties evolve as the DT matures. Finally, we identify the lifecycle phases in which uncertainties become critical. This structured approach facilitates uncertainty management in DTs and is supported by a systematic literature review conducted using the PRISMA methodology.