Automated Analysis of Long-Term Bridge Behavior and Health using a Cyber-Enabled Wireless Monitoring System

A worthy goal for the structural health monitoring field is the creation of a scalable monitoring system architecture that abstracts many of the system details (e. g., sensors, data) from the structure owner with the aim of providing "actionable" information that aids in their decision making process. While a broad array of sensor technologies have emerged, the ability for sensing systems to generate large amounts of data have far outpaced advances in data management and processing. To reverse this trend, this study explores the creation of a cyber-enabled wireless SHM system for highway bridges. The system is designed from the top down by considering the damage mechanisms of concern to bridge owners and then tailoring the sensing and decision support system around those concerns. The enabling element of the proposed system is a powerful data repository system termed SenStore. SenStore is designed to combine sensor data with bridge meta-data (e. g., geometric configuration, material properties, maintenance history, sensor locations, sensor types, inspection history). A wireless sensor network deployed to a bridge autonomously streams its measurement data to SenStore via a 3G cellular connection for storage. SenStore securely exposes the bridge meta-and sensor data to software clients that can process the data to extract information relevant to the decision making process of the bridge owner. To validate the proposed cyber-enable SHM system, the system is implemented on the Telegraph Road Bridge (Monroe, MI). The Telegraph Road Bridge is a traditional steel girder-concrete deck composite bridge located along a heavily travelled corridor in the Detroit metropolitan area. A permanent wireless sensor network has been installed to measure bridge accelerations, strains and temperatures. System identification and damage detection algorithms are created to automatically mine bridge response data stored in SenStore over an 18-month period. Tools like Gaussian Process (GP) regression are used to predict changes in the bridge behavior as a function of environmental parameters. Based on these analyses, pertinent behavioral information relevant to bridge management is autonomously extracted.