The invention provides a track anomaly detection system, which is used for detecting whether a generated track between a starting point and ending point pair in a road network is abnormal or not, the generated track is a track matched with road network information, and the system comprises a track likelihood estimation module which is configured to generate a mode, the likelihood estimation module is used for performing likelihood estimation on the generated trajectory by adopting variational inference to obtain a first score; the depolarization scaling estimation module is configured to generate a model and is used for carrying out depolarization estimation on the generated track by adopting variational inference to obtain a second score; and the abnormity judgment module is used for calculating a track abnormity score according to the first score and the second score of the generated track and a preset rule. According to the method, anomaly detection is decomposed into trajectory likelihood estimation and depolarization scaling estimation, the causal relationship between the starting and ending point pairs and the trajectory is fully modeled, modeling is carried out on the influence of road section preference, depolarization is completed with very low time overhead, the trajectory which is being generated can be efficiently processed, and online detection tasks are supported.

    本发明提供一种轨迹异常检测系统,用于检测路网中起止点对之间已发生轨迹是否异常,所述已发生轨迹是与路网信息经过匹配的轨迹,所述系统包括:轨迹似然估计模块,其被配置为生成模式,用于采用变分推断对已发生轨迹进行似然估计获得第一得分;去偏缩放估计模块,其被配置为生成模型,用于采用变分推断对已发生轨迹进行去偏估计获得第二得分;异常判定模块,用于根据已发生轨迹的第一得分和第二得分按照预设的规则计算轨迹异常得分。本发明将异常检测分解为轨迹似然估计和去偏缩放估,充分建模了起止点对与轨迹之间的因果关系并对路段偏好的影响进行了建模,以非常低的时间开销完成去偏,能高效地处理正在生成的轨迹,支持在线检测任务。


    Zugriff

    Download


    Exportieren, teilen und zitieren



    Titel :

    Trajectory anomaly detection system and online trajectory anomaly detection method


    Weitere Titelangaben:

    一种轨迹异常检测系统、在线轨迹异常检测方法


    Beteiligte:
    LI WENBIN (Autor:in) / YAO DI (Autor:in) / BI JINGPING (Autor:in)

    Erscheinungsdatum :

    2024-01-19


    Medientyp :

    Patent


    Format :

    Elektronische Ressource


    Sprache :

    Chinesisch


    Klassifikation :

    IPC:    G08G Anlagen zur Steuerung, Regelung oder Überwachung des Verkehrs , TRAFFIC CONTROL SYSTEMS / G06N COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS , Rechnersysteme, basierend auf spezifischen Rechenmodellen



    Similarity Based Vehicle Trajectory Clustering and Anomaly Detection

    Fu, Z. / Hu, W. / Tan, T. | British Library Conference Proceedings | 2005


    Similarity based vehicle trajectory clustering and anomaly detection

    Zhouyu Fu, / Weiming Hu, / Tieniu Tan, | IEEE | 2005


    Dynamic scene modelling and anomaly detection based on trajectory analysis

    Wan, Yiwen / Yang, Tze-I / Keathly, David et al. | IET | 2014

    Freier Zugriff

    AI-Empowered Trajectory Anomaly Detection and Classification in 6G-V2X

    Raja, Gunasekaran / Begum, Mubeena / Gurumoorthy, Sugeerthi et al. | IEEE | 2023


    Dynamic scene modelling and anomaly detection based on trajectory analysis

    Wan, Yiwen / Yang, Tze‐I / Keathly, David et al. | Wiley | 2014

    Freier Zugriff