Bridge damage evaluation model based on vehicle excitation and ML

To realize rapid damage assessment of old bridges in remote areas, this paper proposes a rapid damage evaluation platform based on the combination of dynamic response of bridges and Machine Learning (ML). Selecting several typical old bridges, the acceleration signals of these superstructures under different vehicle loads were collected. The damage indicator defines the health status of superstructure, which is developed from the wavelet packet theory. Finally, a set of damage recognition models with universal applicability is established according to the optimized algorithms derived from ML training. The results shows that the proposed approach provides a rapid and economic damage identification of bridges. The approach is a prospective solution for rapid structural evaluation of similar girder bridges.