A Hybrid Method for Vibration-Based Bridge Damage Detection

Abstract

Damage detection algorithms employing the conventional acceleration measurements and the associated modal features may underperform due to the limited number of sensors used in the monitoring and the smoothing effect of spline functions used to increase the spatial resolution. The effectiveness of such algorithms could be increased if a more accurate estimate of mode shapes were provided. This study presents a hybrid structural health monitoring method for vibration-based damage detection of bridge-type structures. The proposed method is based on the fusion of data from conventional accelerometers and computer vision-based measurements. The most commonly used mode shape-based damage measures, namely, the mode shape curvature method, the modal strain energy method, and the modal flexibility method, are used for damage detection. The accuracy of these parameters used together with the conventional sparse sensor setups and the proposed hybrid approach is investigated in numerical case studies, with damage scenarios simulated on a simply-supported bridge. The simulations involve measuring the acceleration response of the bridge to ambient vibrations and train crossings and then processing the data to determine the modal frequencies and mode shapes. The efficiency and accuracy of the proposed hybrid health monitoring methodology are demonstrated in case studies involving scenarios in which conventional acceleration measurements fail to detect and locate damage. The robustness of the proposed method against various levels of noise is shown as well. In the studies considered, damage as small as 10% decrease in flexural stiffness of the bridge and localized in less than 1% of the span-length of the bridge is reliably detected even with very high levels of measurement noise. Finally, a modified modal flexibility damage parameter is derived and used to alleviate the shortcomings of the modal flexibility damage parameter.

Keywords

• damage detection
• vibration-based
• structural health monitoring
• computer vision
• curvature
• strain energy