The invention relates to a method for detecting abnormal behaviors of vehicles in multi-type mixed traffic flow, which comprises the following steps of: acquiring traffic flow information on a road, and abstracting the traffic flow information into a dynamic heterogeneous graph sequence of discrete time according to internal characteristics of the traffic flow information; a long-short-term memory network model using a heterogeneous graph attention module is designed, and time sequence feature extraction of various node problems is realized; and designing an auto-encoder based on a long short-term memory network model, and detecting to obtain an outlier abnormal behavior which does not conform to a normal vehicle decision making mode through a node reconstruction error. Compared with the prior art, the method uses the heterogeneous graph network to learn the vehicle individual information and the interaction behaviors between the vehicles at the same time, can effectively detect the abnormal behaviors of the vehicles, and supports and expands to a large road scene with thousands of vehicles. Therefore, different types of vehicle abnormal behaviors in the mixed traffic flow of various types of vehicles can be rapidly and accurately detected, and safe and efficient operation of a road traffic system is reliably guaranteed.

    本发明涉及一种多种类混合车流中车辆异常行为检测方法,包括以下步骤:获取道路上的车流信息,根据其内部特征,将其抽象为离散时间的动态异构图序列;设计使用了异构图注意力模块的长短期记忆网络模型,实现多种类节点问题的时序特征提取;基于长短期记忆网络模型,设计自编码器,通过节点重构误差,检测得到不符合正常车辆决策方式的离群异常行为。与现有技术相比,本发明使用异构图网络同时对车辆个体信息和车辆间的交互行为进行学习,能够有效地对车辆异常行为进行检测,同时支持扩展到拥有数千车辆级别的大型公路场景,从而针对多种类车辆混合车流中不同种类的车辆异常行为,能够快速、准确地进行检测,可靠保障道路交通系统安全高效的运行。


    Access

    Download


    Export, share and cite



    Title :

    Vehicle abnormal behavior detection method in multi-type mixed traffic flow


    Additional title:

    一种多种类混合车流中车辆异常行为检测方法


    Contributors:
    WANG LIN (author) / LI YUNQING (author) / ZHOU JUNJIE (author)

    Publication date :

    2024-05-14


    Type of media :

    Patent


    Type of material :

    Electronic Resource


    Language :

    Chinese


    Classification :

    IPC:    G08G Anlagen zur Steuerung, Regelung oder Überwachung des Verkehrs , TRAFFIC CONTROL SYSTEMS / G06F ELECTRIC DIGITAL DATA PROCESSING , Elektrische digitale Datenverarbeitung / G06N COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS , Rechnersysteme, basierend auf spezifischen Rechenmodellen / G06Q Datenverarbeitungssysteme oder -verfahren, besonders angepasst an verwaltungstechnische, geschäftliche, finanzielle oder betriebswirtschaftliche Zwecke, sowie an geschäftsbezogene Überwachungs- oder Voraussagezwecke , DATA PROCESSING SYSTEMS OR METHODS, SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL, SUPERVISORY OR FORECASTING PURPOSES




    Towards Detection of Abnormal Vehicle Behavior Using Traffic Cameras

    Wang, Chen / Musaev, Aibek / Sheinidashtegol, Pezhman et al. | British Library Conference Proceedings | 2019


    Traffic abnormal flow causal detection method and device

    SONG XUAN / GAO HAOTIAN / FAN ZIPEI et al. | European Patent Office | 2023

    Free access

    Multi-mode integrated traffic abnormal event detection method

    LI JINGLIN / LUO GUIYANG / YUAN QUAN et al. | European Patent Office | 2024

    Free access

    General modeling method for mixed traffic flow vehicle following behavior

    ZHANG JIAN / WANG XINKE | European Patent Office | 2024

    Free access