Abstract: By means of one-dimension renewal random counting field,the vehicles on highway bridges were de- scribed,and an exponential distribution was derived theoretically to describe the statistic characteristic of travel spacing of vehicles in normal traffic state,which was then verified by the statistic results obtained by weigh-in-mo- tion in situ of the vehicles running naturally on highway bridges,finding that the exponential distribution property can be used to appraise the adaptability of bridges to traffic flow.Then,by taking 3 seconds as a time interval de- marcation,the actually-surveyed vehicles were divided into two running states,namely normal traffic state and dense traffic state,and the difference in time interval and travel spacing of vehicles in these two running states be- tween the statistic results obtained when revising the current bridge design code and the actual results was analyzed comparatively,finding that 3 seconds is not a suitable demarcation of vehicle running state and that travel spacing is more objective and reasonable for the demarcation.By comparing the probability distribution of the travel spacing between actual naturally vehicles on highway bridges and the theoretically-derived vehicles in normal traffic state，it is found that 44.35m is a suitable travel spacing for the demarcation of running state.When the travel spacing is less than 44.35m,vehicles run in dense traffic state and their probability distribution can be regressed as a polyno- mial function,while when the travel spacing is more than 44.35m,vehicles run in normal traffic state and the probability distribution function can be regressed as an exponential one.Such achievements help further establish a multi-parameter load model of random vehicles running on highway bridges.

Key words: highway bridge, travel spacing, one-dimension renewal random counting field, exponential distribu- tion, polynomial regression, normal traffic state, dense traffic state