The invention relates to an abnormal vehicle trajectory anomaly detection method and system based on deep learning, and the method comprises the steps: carrying out the grid processing of obtained vehicle trajectory data, and extracting feature vectors of the vehicle trajectory data after grid processing based on GRU and VAE; the extracted feature vectors are input into a constructed GRU-WGAN model, the GRU-WGAN model comprises a generator and a discriminator, and the generator and the discriminator are both composed of a GRU network; the generator reconstructs track feature points according to input feature vectors, the rebuilt track feature points and real data are input into the discriminator, the discriminator judges whether the track is abnormal or not, parameters of the generator and the discriminator are updated according to the judgment result, output of the training feature extraction part and potential features of the real data are learned, and the real data is extracted. The optimal output result of the discriminator is obtained, and vehicle trajectory data anomaly detection is completed. According to the method, the detection accuracy can be effectively improved, and the interpretability, stability and effectiveness of the model are improved.

    本发明涉及一种基于深度学习的异常车辆轨迹异常检测方法及系统,其包括:将获取的车辆轨迹数据进行网格化处理,并对网格化处理后的车辆轨迹数据基于GRU和VAE提取特征向量;将提取到的特征向量输入到构建的GRU‑WGAN模型中,GRU‑WGAN模型包括一个生成器和一个判别器,生成器和判别器均由GRU网络构成;生成器根据输入的特征向量重建轨迹特征点,重建的轨迹特征点与真实数据输入判别器中,由判别器判定轨迹是否异常,并根据判定结果更新生成器和判别器的参数,学习训练特征提取部分的输出和真实数据的潜在特征,得到判别器最优的输出结果,完成车辆轨迹数据异常检测。本发明能有效提高检测准确度,并提高了模型的可解释性和稳定性、有效性。


    Access

    Download


    Export, share and cite



    Title :

    Abnormal vehicle track anomaly detection method and system based on deep learning


    Additional title:

    一种基于深度学习的异常车辆轨迹异常检测方法及系统


    Contributors:
    WANG LEI (author) / LIU YUHANG (author) / ZHAO XIAOYONG (author) / CUI GUOXI (author) / ZHANG JINGLE (author) / WANG NINGNING (author) / LU HUIYA (author)

    Publication date :

    2023-07-11


    Type of media :

    Patent


    Type of material :

    Electronic Resource


    Language :

    Chinese


    Classification :

    IPC:    G06F ELECTRIC DIGITAL DATA PROCESSING , Elektrische digitale Datenverarbeitung / G06N COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS , Rechnersysteme, basierend auf spezifischen Rechenmodellen / G08G Anlagen zur Steuerung, Regelung oder Überwachung des Verkehrs , TRAFFIC CONTROL SYSTEMS



    Vehicle track detection method based on deep learning

    ZHANG ZHONG / SUN GUANLONG / HUANG JUNJIE | European Patent Office | 2022

    Free access


    ABNORMAL VEHICLE DETECTION DEVICE, ABNORMAL VEHICLE DETECTION SYSTEM, ABNORMAL VEHICLE DETECTION METHOD, AND PROGRAM

    SHIRATORI YUSUKE / NIWA EIJI / MORI YUHEI | European Patent Office | 2021

    Free access

    RAILROAD TRACK ANOMALY DETECTION

    LI SAISHI FRANK | European Patent Office | 2019

    Free access

    Rail surface anomaly detection based on deep learning

    Shi, Lei / Wu, Junjie / Sun, Yongkui | SPIE | 2023