面向实时交通信息提取的车辆轨迹数据挖掘

Trajectory Mining for Live Traffic Condition Retrieving

车辆行驶轨迹是驾驶人员主观意愿和道路客观约束条件综合作用的结果,从海量车辆轨迹中可以挖掘出道路的实时交通信息为智能交通服务.通过建立轨迹约束模型用以量化各约束因子,基于线性参照系统的数据预处理以加快检索速度和降低轨迹的不确定性,基于移动目标主体相似性和移动轨迹时空相似性的数据选取降低了数据库搜索次数提高发掘准确度,分别针对道路交叉口和一般路段进行数据挖掘,提取实时的道路交通信息.研究结果和实验结果表明,这个挖掘算法可以有效地提取道路实时路况信息.

Trajectories, as the synthetic results of moving entity＇s purposes and environment con- straints, can be the mining resources oriented to live traffic condition retrieving for intelligent traffic applications. This paper proposes a trajectory mining method consisting of constraints modeling to quantify variant sub-constraints, data preprocessing based on linear reference system to increase indexing efficiency and decrease trajectories uncertainty, data selecting based on moving entity similarity and spatio-temporal similarity to decrease searching times and increase mining accuracy, and two mining algorithms being respectively implemented near road intersection and general road segment away from road intersection. Its efficiency is verified by investigation and experiment result.