The invention relates to an urban road accident congestion two-stage detection method based on trajectory data. The method comprises two stages: abnormal trajectory fragment identification and accident congestion detection. In the abnormal trajectory segment identification stage, two key features (average speed and time for entering a road section) are defined to capture trajectory segment features in a space-time window, and the features are jointly influenced by accident congestion and downstream intersection signal timing. And on the basis of the two key features, abnormal track fragments are identified through clustering. In an accident congestion detection stage, an abnormal rate of a space-time window is defined. And based on the abnormal rate, identifying an accident congestion space-time window by using a decision tree. Compared with the prior art, the method is suitable for the urban road network, and achieves the accident congestion detection of the signal period level in time and the vehicle flow direction level in space.
本发明涉及一种基于轨迹数据的城市道路事故性拥堵两阶段检测方法,包含两个阶段:异常轨迹片段识别、事故性拥堵检测。在异常轨迹片段识别阶段,定义了两个关键特征(即平均速度和进入路段的时间)来捕捉时空窗内的轨迹片段特征,该特征受事故性拥堵和下游交叉口信号配时的共同影响。基于这两个关键特征,通过聚类对异常轨迹片段进行识别。在事故性拥堵检测阶段,定义了时空窗口的异常率。基于异常率,利用决策树识别事故性拥堵时空窗。与现有技术相比,本发明适用于城市路网,实现了时间上信号周期级、空间上车辆流向级的事故性拥堵检测。
Urban road accident congestion two-stage detection method based on trajectory data
一种基于轨迹数据的城市道路事故性拥堵两阶段检测方法
2025-04-11
Patent
Electronic Resource
Chinese
IPC: | G08G Anlagen zur Steuerung, Regelung oder Überwachung des Verkehrs , TRAFFIC CONTROL SYSTEMS |
Analysis of GPS Based Vehicle Trajectory Data for Road Traffic Congestion Learning
Springer Verlag | 2014
|European Patent Office | 2024
|