基于车辆自动识别技术的动态OD矩阵估计新方法

A Novel Dynamic OD Estimation Approach Based On Automatic Vehicle Identification Data

摘要:

根据车辆自动识别（Automatic Vehicle Identification，AVI）技术检测的新信息，提出了利用AVI检测信息进行动态OD估计的新方法。该方法以粒子滤波算法思想为基础，引入AVI检测的部分路径信息、动态行程时间信息以及检测器可测性判据，首先通过贝叶斯估计算法对路径缺失的车辆信息进行缺失路径范围确定和选择概率修正；然后利用蒙特卡洛随机仿真模拟任意车辆对缺失路径的选择，进而获得基于个体车辆部分路径的初始修复OD矩阵；最后运用AVI检测的流量信息校正初始修复OD矩阵，得到最终的OD矩阵估计值。以上海市南北高架快速路网为研究对象，分析了在不同AVI覆盖率和先验精度条件下的动态OD估计精度，结果表明在60 %及以上的AVI覆盖率都能获取相当高的OD估计精度; 即使在50%覆盖率和60%的先验精度条件下运用本方法OD估计的平均相对误差仍为28.87%。同时本方法对OD先验信息的精度要求较低，能更好的满足我国目前OD基础数据精度不高的现实需求。

Abstract:

Based on the new information which was detected through Automatic Vehicle Identification (AVI) technology, this paper proposed a new approach for dynamic OD estimation by using the AVI information. Partial trajectory information, dynamic travel time information and detector measurability were introduced into this approach which refer to the particle filter. First the selection scope and the selection probability were reduced and collected by Bayesian Estimation. Then the absented trajectory of any vehicles was determined by Monte Carlo stochastic simulation and the initial corrected OD matrix was obtained by correcting the individual vehicles trajectory. At last, the initial OD matrix was corrected by the path-link flow function based on the AVI volume information. Finally, the paper analyzed the accuracy of dynamic OD estimation on different coverage of AVI and different accuracy of prior information based on the Shanghai North-South expressway simulation model, the result showed that the accuracy of OD estimation was high when the coverage was more than 60% and relative error of dynamic OD estimation was 28.87% in 50% coverage and 60% accuracy of prior information. It also showed that this approach could be used with low accuracy prior information which can better meet the field need that the current OD information is low precision in China.