Phy-MFN: Physics-informed Multimodal Fusion Network for Bridge Crack Evolution Prediction

Résumé fourni par la source

This record provides the proof-stage open research package for the Phy-MFN framework proposed in the manuscript "Physics-informed Multimodal Fusion Network for Robust Early-stage Crack Evolution Prediction in Long-span Concrete Bridges under Climatic Thermal Drifts". The package includes the Phy-MFN algorithm scaffold, physics-informed loss functions, adaptive wavelet preprocessing utilities, data de-identification scripts, a data dictionary, and release documentation. This release does not include raw bridge monitoring data, bridge-identifying information, exact sensor coordinates, or raw UAV imagery. A de-identified version of the 45-day monitoring sequence will be deposited after institutional and bridge-authority approval.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.