It has been almost two decades since the studies on bridge modal identification using scanning vehicles (i.e., vehicle instrumented with sensor(s)) started. The underlying principle is that, when a scanning vehicle traverses a bridge, the measured response contains the bridge response [53]. Therefore, it is possible to retrieve information of the bridge indirectly using the vehicle response, hence derives the name indirect bridge identification. Compared to conventional identification which uses permanently-installed stationary sensor network, the indirect approach entails the advantages of reduced cost and higher mobility. However, indirectly identifying the bridge modal properties is not as straight-forward as the conventional method. This is primarily due to the mixed space-t...[ Read more ]

It has been almost two decades since the studies on bridge modal identification using scanning vehicles (i.e., vehicle instrumented with sensor(s)) started. The underlying principle is that, when a scanning vehicle traverses a bridge, the measured response contains the bridge response [53]. Therefore, it is possible to retrieve information of the bridge indirectly using the vehicle response, hence derives the name indirect bridge identification. Compared to conventional identification which uses permanently-installed stationary sensor network, the indirect approach entails the advantages of reduced cost and higher mobility. However, indirectly identifying the bridge modal properties is not as straight-forward as the conventional method. This is primarily due to the mixed space-time nature of the vehicle-bridge-interaction system. In addition, since the indirect approach uses response measured by a single traversing vehicle, this also means significantly less data is available for identification purposes.

This study proposes a systematic framework to extract the bridge dynamic characteristics from response data collected by multiple traversing vehicles. The basis of the method is to form a response matrix, which is information-rich in terms of space and time, with multiple scanning vehicle response measurements. This circumvents problems caused by the space-time-varying nature of the vehicle-bridge system. The novelty of the study lies in a generalized contact point response estimation, as well as , for the first time, the introduction of a smoothness regularizer into a objective function for matrix completion algorithm. This study considers a vehicle-bridge-interaction system subjected to a stochastic highway traffic flow, modeled by cellular automata. Identification results based on the numerically simulated response demonstrates that the proposed method can estimate the bridge natural frequencies and mode shapes with significant accuracy.

keyword: vehicle-bridge interaction, modal identification, matrix completion, crowd-sensing