Statistical Estimation of Dynamic Travel Time and Traffic Flow Propagation

초록

This paper formulates dynamic travel time models under density-speed equilibrium states through a statistical approach. The dynamic travel time models are formulated by assuming that travel times are determined by the distribution of the traffic stream condition with respect to either location or time. Then, this paper shows that traffic propagations can be explicitly expressed by a function of a dynamic travel time model. In density-speed equilibrium models,such as the LWR model and the proposed dynamic travel time models, the vehicular speed is instantly adjusted to the density, regardless of the speed of the subject vehicle. As a result, vehicular accelerations in density-speed equilibrium models do not always fall within a feasible region of the vehicular performance. In order to formulate the actual traffic behaviors regarding vehicular accelerations, this paper proposes a non-equilibrium traffic flow model (i.e., continuum car-following model).