EVALUATION OF GIRDER STIFFNESS FLUCTUATION DUE TO CRACK BREATHING UNDER TRAIN PASSAGES USING BAYESIAN TV-ARX MODEL

Abstract

Cracks in PRC girders are important index related to girder stiffness. However they are often closed except loading timing due to the introduced prestress, thus, it is difficult to detect visually. In the vibrationbased evaluation, it is necessary to separate the exciation characteristics and then evaluate the rapid fluctuation of the bridge frequency depending on the amplitude. In this study, the Bayesian TV-ARX method, which is a time-varying system identification method considering external force characteristics, are applied to a measured displacement response of high-speed railway PRC girder when a train passage. It was founded that the instantaneous decrease of bridge frequency that cannot be estimated by the continuous Wavelet transform occurred in the target bridge. In addition, crack opening at the time of the lower amplitude is a main cause and it can reduce the girder bending stiffness by a maximum of approximately 30% during train passage.