Abstract
The design of smart, autonomous mechanical structures capable of performing real-time monitoring of their integrity and taking proactive measures during operation to prevent downtime or failure has become an active area of research. There is a critical need in various industrial sectors (transportation, energy, etc.) for greater reliability, as well as improved performance and durability of equipment (aircraft, wind turbines, bridges, etc.). Implementing such advanced technology would enable optimized maintenance and the ability to operate in degraded mode, managing the reduction in load-carrying capacity by adapting the operating plan.
However, real-time monitoring of damage in engineering systems—achieved by dynamically coupling predictive simulation tools (such as digital twins) with sensor observations—is extremely challenging in practice due to several issues. In particular, the complex nonlinear multiscale phenomena involved may require computationally intensive simulations (which are hardly compatible with real-time processing), necessitating reduced-order modeling and effective strategies for data assimilation and control. In addition, the problem is plagued by model bias, environmental uncertainty, and measurement noise, all of which must be taken into account for accurate diagnosis and prognosis, as well as safe decision-making. In this context, a promising trend is the use of hybrid twins, in which an a priori physics-guided model is updated and enriched on the fly with data-driven information, thereby leveraging all available knowledge.
