Validation of vision-based vibration measurements for finite element modal updating of a cable-stayed bridge

This study presents a finite element (FE) model updating approach for a laboratory-scale cable-stayed bridge using vibration data obtained from a vision-based measurement system. Conventional accelerometer-based responses were used as a reference for comparison. The bridge’s dynamic properties were extracted through experimental modal analysis under controlled excitation using both pseudo-random and sine-chirp signals. Optical motion tracking was performed using Template Matching (TM) and Optical Flow (OF) techniques, with the latter demonstrating superior accuracy and stability. A sensitivity-based, iterative Bayesian updating framework was employed to adjust key physical parameters, including the moduli of elasticity of the stay cables and stiffness at the deck supports. Results show excellent agreement between updated model predictions and experimental data, with modal frequency errors reduced from up to 16% to less than 1.5% and Modal Assurance Criterion (MAC) values exceeding 99%. The outcomes confirm that camera-based vibration monitoring can provide reliable data for FE model updating, achieving comparable accuracy to accelerometer based methods while maintaining the advantages of non-contact and easily deployable measurements.