By seamlessly integrating a reduced-order digital twin containing damage states and influencing parameters as a decoder within the variational neural network，it enables the individualized, real-time structural damage quantification and parameter calibration across an entire fleet, while accounting for uncertainties.
