The diagram illustrates a physics-informed digital twin architecture for polymer aging. It includes governing degradation physics such as reaction-diffusion, Arrhenius kinetics, and property-damage coupling embedded within a Physics-Informed Neural Network (PINN). Short-term thermal-oxidative aging data and physical constraints are jointly assimilated to learn material parameters and generate stable long-term degradation predictions within a closed feedback loop. The diagram also compares classical PDE solvers, purely data-driven machine-learning models, and the proposed PINN framework in terms of data requirements, physical consistency, extrapolation stability, and parameter learning.Conceptual illustration of the proposed physics-informed digital twin framework and its methodological context. (a) Physics-informed digital twin architecture for polymer aging, in which governing degradation physics reaction–diffusion, Arrhenius kinetics, and property–damage coupling are embedded within a Physics-Informed Neural Network (PINN). Short-term thermal-oxidative aging data and physical constraints are jointly assimilated to learn material parameters (e.g. activation energy and diffusion coefficient) and to generate stable long-term (10-year) degradation predictions within a closed feedback loop. (b) Conceptual comparison of modeling approaches, contrasting classical PDE solvers, purely data-driven machine-learning models, and the proposed PINN framework in terms of data requirements, physical consistency, extrapolation stability, and parameter learning, highlighting the PINN as a balanced and physically consistent approach for digital-twin applications