Modern engineering systems face complex, nonlinear, and dynamic conditions that challenge traditional modeling and monitoring approaches. While advances in sensing technologies have enabled detailed data collection, these datasets are often sparse, noisy, and distorted by environmental and operational factors. As a result, purely data-driven methods often struggle to effectively capture system behavior. This talk presents a physics-enhanced digital twin framework that combines physics-based understanding - particularly of system dynamics - with advanced data assimilation techniques. By integrating domain knowledge with observational data, these augmented twins provide interpretable, robust, and scalable solutions for system management. Such a hybrid approach overcomes key limitations of traditional methods and supports improved decision-making for complex, real-world systems. The presentation underscores how embedding physical principles into digital models enables more resilient, aware and responsive engineering infrastructure.
Eleni Chatzi (Mon,) studied this question.