A Cross-Simulation Method for Large-Scale Traffic Evacuation with Big Data

摘要

Microscopic traffic simulation is one of the effective tools for transportation forecast and decision support. It is a challenge task to make reasonable prediction of traffic scenarios during emergency. Big data technology provides a new solution for this issue. This paper proposes a cross-simulation method to apply the mass data collected in normal situations into large-scale traffic evacuations to provide better supporting information for emergency decision. The method consists of three processes: Acquisition, Analysis and Adaptation. It captures the dynamic distance-speed relation of every vehicles on the real roads and build a database of driving behaviors according to the existing car-following models. After calibration and analysis, various driving behaviors can be identified. During emergency, the distribution of driving behaviors will be refactored to adapt the fast-changing situation automatically so that the simulation system gains the adaptive ability in emergency situations. An experimental result on a real road preliminarily validates the practicability of the method and shows the supporting information which it can provide. The new method will make contributions on enhancing the predictive ability of traffic simulation systems in emergency situations.