基于聚类的出租车异常轨迹检测
发布时间：2018-09-06 12:44
【摘要】：出租车全球定位系统数据中蕴含城市交通和移动对象行为的宏观信息,从中可以挖掘出有价值的异常轨迹模式。将位置和几何形状、行驶时间分别作为出租车轨迹的空间与时间特征,根据特征偏离情况划分时间、空间和时空异常轨迹。从轨迹数据中提取相同起终点的轨迹集,将轨迹划分成轨迹片段,计算轨迹间的相似度并进行基于距离和密度的聚类,在空间特征上初步分离出频繁和稀疏轨迹,根据数据异常判定的kσ准则确定时间特征异常的分离阈值,对时间特征进行再次划分,最终实现出租车异常轨迹检测。实验结果表明,该方法能从异常轨迹中挖掘出个性化路线、异常停留位置和交通路段,为智能交通、物流高效规划和执行等提供参考信息。
[Abstract]:The global positioning system (GPS) data of taxis contain macro information about the behavior of urban traffic and moving objects, from which valuable abnormal trajectory patterns can be mined. The position, geometric shape and travel time are regarded as the space and time characteristics of the taxi track respectively, and the time, space and time abnormal track are divided according to the characteristic deviation. The trace sets of the same beginning and end point are extracted from the trajectory data, the trajectory is divided into trajectory segments, the similarity between the tracks is calculated and clustering based on distance and density is carried out, and the frequent and sparse trajectories are preliminarily separated from the spatial features. According to the k 蟽 criterion of data anomaly determination, the separation threshold of time feature anomaly is determined, and the time feature is divided again, finally the taxi abnormal track detection is realized. The experimental results show that the method can mine personalized routes, abnormal stop positions and traffic sections from abnormal tracks, and provide reference information for intelligent transportation, efficient planning and execution of logistics, etc.
【作者单位】： 信息工程大学地理空间信息学院;
【基金】：国家自然科学基金“空间数据流的概念漂移问题研究”(41571394)
【分类号】：U495;TP311.13
，
本文编号：2226391
[Abstract]:The global positioning system (GPS) data of taxis contain macro information about the behavior of urban traffic and moving objects, from which valuable abnormal trajectory patterns can be mined. The position, geometric shape and travel time are regarded as the space and time characteristics of the taxi track respectively, and the time, space and time abnormal track are divided according to the characteristic deviation. The trace sets of the same beginning and end point are extracted from the trajectory data, the trajectory is divided into trajectory segments, the similarity between the tracks is calculated and clustering based on distance and density is carried out, and the frequent and sparse trajectories are preliminarily separated from the spatial features. According to the k 蟽 criterion of data anomaly determination, the separation threshold of time feature anomaly is determined, and the time feature is divided again, finally the taxi abnormal track detection is realized. The experimental results show that the method can mine personalized routes, abnormal stop positions and traffic sections from abnormal tracks, and provide reference information for intelligent transportation, efficient planning and execution of logistics, etc.
【作者单位】： 信息工程大学地理空间信息学院;
【基金】：国家自然科学基金“空间数据流的概念漂移问题研究”(41571394)
【分类号】：U495;TP311.13
，
本文编号：2226391
本文链接：https://www.wllwen.com/kejilunwen/ruanjiangongchenglunwen/2226391.html
