Wireless Monitoring Systems for Long-Term Reliability Assessment of Bridge Structures based on Compressed Sensing and Data-Driven Interrogation Methods.

Abstract

The state of the nation’s highway bridges has garnered significant public attention due to large inventories of aging assets and insufficient funds for repair. Current management methods are based on visual inspections that have many known limitations including reliance on surface evidence of deterioration and subjectivity introduced by trained inspectors. To address the limitations of current inspection practice, structural health monitoring (SHM) systems can be used to provide quantitative measures of structural behavior and an objective basis for condition assessment. SHM systems are intended to be a cost effective monitoring technology that also automates the processing of data to characterize damage and provide decision information to asset managers. Unfortunately, this realization of SHM systems does not currently exist. In order for SHM to be realized as a decision support tool for bridge owners engaged in performance- and risk-based asset management, technological hurdles must still be overcome. This thesis focuses on advancing wireless SHM systems. An innovative wireless monitoring system was designed for permanent deployment on bridges in cold northern climates which pose an added challenge as the potential for solar harvesting is reduced and battery charging is slowed. First, efforts advancing energy efficient usage strategies for WSNs were made. With WSN energy consumption proportional to the amount of data transmitted, data reduction strategies are prioritized. A novel data compression paradigm termed compressed sensing is advanced for embedment in a wireless sensor microcontroller. In addition, fatigue monitoring algorithms are embedded for local data processing leading to dramatic data reductions. In the second part of the thesis, a radical top-down design strategy (in contrast to global vibration strategies) for a monitoring system is explored to target specific damage concerns of bridge owners. Data-driven algorithmic approaches are created for statistical performance characterization of long-term bridge response. Statistical process control and reliability index monitoring are advanced as a scalable and autonomous means of transforming data into information relevant to bridge risk management. Validation of the wireless monitoring system architecture is made using the Telegraph Road Bridge (Monroe, Michigan), a multi-girder short-span highway bridge that represents a major fraction of the U.S. national inventory.