Abstract

A monitoring strategy for investigating the long-term longitudinal deflections at the expansion joints of the I-35W St. Anthony Falls Bridge was developed as a means of inferring the integrity of the bearings. During the duration of this study, the deflections were primarily caused by time-dependent behavior, which tends to be highly uncertain, and changes in temperature, which are comparatively predictable. The proposed monitoring strategy uses separate anomaly-detection routines to monitor the expansion joints for quickly developing problems, such as bearing lockup, and slowly developing problems, such as unexpected changes in the rate of time-dependent bearing movement that might be indicative of slow degradation. The quickly developing anomaly-detection routine presented used Bayesian regression to combine the uncertainty in the time-dependent deflection predictions with the scatter of the data to realize coherent bounds for discerning problems that occur instantaneously or over the course of several weeks. The slowly developing anomaly-detection routine used the rates of the measured time-dependent behavior with respect to the Arrhenius-adjusted age to discern deterioration developing over time frames from several months up to several years. These rates were expected to decrease with time according to a power function for a structure with no damage. The system was tested on the collected linear potentiometer data from the I-35W St. Anthony Falls Bridge to investigate the incidence of false positives. Use of model code time-dependent provisions for the two anomaly-detection routines minimized false positives. To test the efficacy of the developed system in identifying true positives, artificial perturbations were introduced to the collected data. These perturbations were intended to mimic probable quickly developing and slowly developing problems that the bridge might encounter. The system successfully identified the introduced quickly developing and slowly developing perturbations unless sensor failure followed by a delayed sensor replacement occurred at approximately the same time as introduction of the perturbation.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call