基于高速公路收费系统的车辆全路径贝叶斯推断匹配技术研究

Research on Bayesian Inference Method Matching Approach of Vehicle Full Route Based on Highway Toll System Data

利用高速公路通行费收费系统监测数据,针对现行高速公路通行费分省联网的收费数据无法真实反映跨省车辆完整行驶路径的问题,引入了车牌串接贝叶斯推断匹配技术。通过对现有收费系统主要指标数据可利用性和逻辑关联性分析,最终选择了通行时间、车种、车辆轴数、车货总重等属性指标作为匹配推断的参考指标,设计了匹配方法和具体流程,对现行高速公路通行费收费系统中跨省通行车辆的完整行驶路径进行了车牌匹配和全路径模拟还原。并选择了2个典型省界收费站进行了实证分析,测算结果表明利用推荐的匹配算法可以把货运车车辆的匹配度提高20个百分点。

In this paper monitored data of tolling system for expressway is used,in regard with the problem that tolling data from current tolling network system by provinces cannot truthfully reflect the complete travel path of trans-provincial vehicles, the Bayesian Inference Method is brought in. Through analysis on availability and logical relevance of main indicator data from current toll system, travel time, vehicle type, vehicle axle numbers, car cargo gross weight, etc. are selected as attribute indicators for matching inference. Specific matching method and process are designed, and vehicle identification and full route restoration simulation are conducted on vehicles trans-provinces in current highway toll system. Two representative trans-provincial toll stations are selected to conduct empirical analysis. The results show that matching rate of freight increases by 20 percent by means of the recommended matching algorithm.