Answers are grounded in this paper and cite their sources.
Vibration-based structural damage identification (SDI) often relies on accelerometers, but their dense deployment is constrained by installation complexity, limiting the spatial richness of the available modal information. Visual sensing provides non-contact displacement measurements with high spatial coverage. To exploit their complementary measurement strengths, this study proposes a finite element model updating (FEMU)-based SDI framework. In this framework, a cascaded visual sensing strategy combines deep learning-based target tracking with optical flow to extract sub-pixel displacement responses from structural videos. The extracted displacement responses are used to identify mode shapes and derive modal strain energy. These vision-derived modal features and acceleration-derived natural frequencies are incorporated into a single objective. The resulting non-convex FEMU inverse problem is solved using a hybrid strategy that combines the Snake Optimizer with Sequential Quadratic Programming. The framework is evaluated through two numerical case studies and a laboratory validation on a five-story steel frame. Across these cases, the framework consistently identified the imposed stiffness-reduction locations and produced stable estimates across repeated runs. These results demonstrate the feasibility of acceleration-vision fusion within FEMU for localizing and quantifying parameterized stiffness reductions in frame structures.
No takes yet. Share an insight, caveat, or question.
Zhang et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: