Data interpretation for infrastructure system diagnosis and prognosis based on model falsification

Measurements obtained from sensors enable engineers to evaluate the condition of existing infrastructure systems, thus providing valuable information for asset managers. With the evolution, miniaturization and cost reductions in sensors, data acquisition systems and digital computing hardware, various kinds of measurements and real time monitoring are available. However, interpreting the data that is collected from sensor networks remains a challenge. This thesis focuses on the data interpretation for infrastructure-system diagnosis and prognosis. The infrastructure systems examined in this thesis include train-track systems, vehicular bridges and footbridges. The author has contributed in four significant aspects: (1) enhancing static load test identification of bridges using dynamic data, (2) time-series data interpretation for wheel flat identification including uncertainties, (3) vibration serviceability assessment for pedestrian bridges, and (4) economic benefit of the updated bridge loading capacity in toll highways.