Hybrid Sensor-Camera Monitoring for Damage Detection: Case Study of a Real Bridge

Abstract

This article presents the real-world implementation of a novel monitoring system in which video images and conventional sensor network data are simultaneously analyzed to detect possible damage on a movable bridge. The monitoring system was designed to detect such problems at the onset of damage. A video stream of traffic is processed to detect and classify vehicles to determine the vehicle load and location, while strain measurements are simultaneously collected at various critical locations on the bridge for both normal and damage conditions. A series of unit influence lines can then be extracted for all of the scenarios using the image and sensor data. Because large data sets result from continuous monitoring, the system also includes a statistical outlier-detection algorithm. The proposed methodology was successfully used to detect and locate common damage scenarios on a real-world bascule bridge.