ABSTRACT
Bridge dynamic properties can be extracted through the analyses of crowdsourced smartphone-vehicle trip (SVT) data collected during routine trips over bridges. The key advantage of crowdsourced SVT data is the ease in generating large volumes of data at very low costs compared to modern structural health monitoring (SHM) systems. By default, these SVT datasets generally contain measurements from over a dozen sensors embedded within the smartphone. Such SVT metadata are presumed to capture important variables that influence the inherent quality of the vibration data and the resulting accuracy of dynamical property estimates. However, the quantitative relationships between key metadata variables such as vehicle speed or trip direction are not well understood in real-world scenarios. This paper studies how the estimates of the absolute valued mode shapes (AMS) are affected by metadata variables of smartphone-vehicle trip (SVT) datasets. This study considers SVT data collected on two bridges (i) the Golden Gate Bridge in California, and (ii) the Gene Hartzell Memorial Bridge in Pennsylvania and focuses on the influence of vehicle speed and vehicle travel direction. A better understanding how such variables impact the estimation of bridge dynamic properties is critical to the future design and implementation of effective crowdsourcing campaigns.
