Time-domain modal identification of bridges based on uncertainty quantification

This study aims to propose an efficient modal identification method for bridges under operation. The noisy condition caused by the traffic loadings is one of the difficulties involved in the operational modal identification. To cope with the problem, this study quantifies uncertainty involved in the modal properties utilizing Bayesian statistics. The quantified uncertainty enables to determine the reasonable model order and to extract the stably estimated modal properties from the determined model. The proposed method is applied to traffic induced vibration measured from an actual truss bridge. The extracted modal frequencies well correspond to peaks in power spectral density curves. Twelve bending and torsional modes are efficiently extracted by the proposed method. Six modes in the twelve is possibly newly observed modes.