Deflection is a crucial indicator to reflect the operating condition of girder bridges, which can be used to evaluate structure condition and identify abnormal loading. The paper analyzed the deflection characteristics of long-span girder bridges based on the coupling vibration between stochastic traffic stream and bridge. First, the latest research advances were integrated to form an analytical model of the coupling vibration between stochastic traffic stream and bridge. Then, a generalized Pareto distribution model based on peaks-over-threshold theory was established to predict the extreme girder deflection. Next, a cellular automaton based microsimulation method was proposed to model the traffic loads on bridges, which utilized the intelligent driver car-following model and acceptance distance based lane-changing model. Finally, these theories were applied in the case study of a long-span prestressed concrete continuous girder bridge. It is discovered from the study that, under the coupling vibration between stochastic traffic stream and bridge, the predicted extreme deflection of the case bridge is far lower than the specified design value. Hence, a grading warning model was established and employed to the analysis of deflection monitoring data of the bridge, showing a wide potential prospect of application.
Keywords: cellular automaton; deflection; long-span girder bridge; traffic microsimulation; vehicle bridge interaction.