Damage identification of simply supported bridges based on optical-flow-derived displacement influence lines

To address the limitations of conventional bridge inspection methods, including inefficiency, subjectivity, and limited quantification and localisation, this study proposes a vision-based approach for deformation monitoring and damage identification of simply supported bridges using optical flow tracking from image sequences. A non-contact, quasi-full-field, quantitative, and cost-effective framework for structural condition assessment is developed. To mitigate displacement drift and limited localisation accuracy in conventional optical flow methods, a motion consistency constraint is incorporated into the tracking framework, and an improved optical flow algorithm with discrete feature point correction is proposed through periodic calibration and abnormal flow elimination. Numerical simulations and laboratory experiments show displacement errors within 5% compared with reference sensors, indicating improved accuracy and robustness. Furthermore, based on intact and damaged displacement influence line theory, a damage indicator using the second-order derivative of fused displacement differences is established. Finite element and experimental results show that the proposed indicator can accurately localise damage under both single- and multiple-damage scenarios. The peak amplitude of the indicator increases with damage severity and therefore provides a basis for relative damage severity assessment. The proposed method requires displacement influence lines from a few measurement points, offering a practical solution for vision-based bridge health monitoring.