Digital Twin-Integrated Reduced-Order Modeling Framework for Real-Time System Prediction

Abstract

Digital Twin technologies have become a new potent paradigm in the real-time monitoring and predictive analysis of complicated engineering systems. Its usefulness, however, is usually limited by the prohibitively large computational cost of full-order numerical models, which makes it hard to apply in real-time. To tackle this issue, this paper suggests a Digital Twin-based system employing Reduced-Order Modeling (ROM) to predict the system efficiently and accurately. The suggested solution incorporates data-driven synchronization and a reduced-order representation obtained with the help of the methods, including Proper Orthogonal Decomposition, allowing to perform quick-computation and still capture the most important system dynamics. The framework is tested on benchmark simulations and the predication accuracy is measured by such metrics as Root Mean Squared Error (RMSE) and relative error. Findings show that the suggested approach has a high prediction fidelity, and the errors are within reasonable limits as compared to full-order models. Moreover, the computational efficiency is significantly enhanced as well, and the speedup is quite significant, allowing the real-time or close to it performance. The primary value-added of this work is the smooth interconnection between Digital Twin architecture with reduced-order modeling to offer a computationally efficient solution with scalability to real-time system prediction. This framework has good potentiality in use in engineering systems with a need to have fast and reliable predictive functions, which takes the gap between accuracy and computational feasibility.