Extraction of Bridge Frequencies from a Moving Test Vehicle by Stochastic Subspace Identification

Abstract

The technique for extracting bridge frequencies from a moving vehicle was proposed only recently. However, its efficacy may be reduced by road-surface roughness. To this end, the stochastic subspace identification (SSI) is modified to deal with the present time-variant coupled noisy vehicle–bridge interaction system. First, the governing equations for the vehicle and bridge are expressed in state space, including the effects of road roughness and multiple vehicles. Then, they are discretized and transformed to a form suitable for SSI by separating the known from the unknown parameters. Using the Hankel matrix along with the orthogonal projection theorem and singular value decomposition, the observability matrix derived with the vehicle effect suppressed can be used to identify the bridge dynamic properties. It is demonstrated that (1) the proposed SSI approach is more effective for identifying bridge frequencies below 20 Hz for the cases studied, compared with conventional approaches; (2) adding a little damping to the test vehicle can help suppress the vehicle frequency using the proposed approach; and (3) the ongoing traffic is beneficial for amplifying the frequencies of bridges with rough surface.