The invention discloses a similar node dynamic perception traffic flow prediction method with space-time heterogeneous attention, and the method comprises the following steps: S1, designing a dynamic parameter memory generator, and dynamically generating parameters in a gated attention unit and a space self-attention module according to input features; s2, constructing a similar node sensing space-time incidence matrix based on a Wasserstein distance, and fusing the space-time incidence matrix into a space self-attention module by adopting a graph shielding method; s3, capturing space-time correlation by using a gating attention unit and a space self-attention module respectively; and S4, fusing the spatio-temporal features extracted by the gating attention unit and the spatial self-attention module through a heterogeneous feature fusion module. The invention provides a novel similar node dynamic perception traffic flow prediction model with spatio-temporal heterogeneity attention, and the new model endows node data with higher similarity with larger weight through dynamic parameter adjustment, so that the prediction performance of the model is improved.

    本发明公开了一种带时空异质注意力的相似节点动态感知交通流预测方法,包括以下步骤:S1:设计一个动态参数记忆生成器,按照输入特征动态生成门控注意力单元和空间自注意力模块中的参数;S2:基于Wasserstein距离构建相似节点感知时空关联矩阵,并采用图遮蔽方法将时空关联矩阵融入到空间自注意力模块中;S3:分别使用门控注意力单元和空间自注意力模块来捕获时空相关性;S4:通过一个异构特征融合模块将门控注意力单元和空间自注意力模块提取的时空特征进行融合。本发明提出了一个新型的带时空异质注意力的相似节点动态感知交通流预测模型,新模型通过动态参数调整,对相似度更高的节点数据赋予更大的权重,从而提升模型预测性能。


    Access

    Download


    Export, share and cite



    Title :

    Similar node dynamic perception traffic flow prediction method with space-time heterogeneous attention


    Additional title:

    带时空异质注意力的相似节点动态感知交通流预测方法


    Contributors:
    LOU JUNGANG (author) / YANG JINNAN (author) / CUI WENTIAN (author) / SHEN QING (author) / LIU ZHENFANG (author)

    Publication date :

    2024-08-02


    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



    Traffic flow prediction method based on dynamic space-time similar pyramid network

    WANG XING / CHEN FEIFEI / JIN BIAO et al. | European Patent Office | 2024

    Free access

    Transfer perception-based time-space diagram attention network traffic flow prediction method and system

    ZHOU YAN / WANG XIAODI / JIA JIPENG et al. | European Patent Office | 2025

    Free access

    Vehicle track prediction method based on heterogeneous node time-space domain perception

    GAO MINGYU / ZENG XUEPU / YANG JUNJIE et al. | European Patent Office | 2023

    Free access

    Attention-based dynamic space-time diagram convolution traffic flow prediction system and method

    JI ZHENYUAN / HUAN YUEHUI / YANG GENKE et al. | European Patent Office | 2023

    Free access

    Traffic flow prediction method based on space-time attention network

    MA CHUANG / YAN LI / LIU SHUAIWU et al. | European Patent Office | 2023

    Free access