The invention discloses an urban regional traffic flow prediction method considering multiple space-time granularities. The method comprises the following steps: firstly, acquiring a traffic flow data set, a regional data set, a weather data set and a point of interest (POI) data set; secondly, preprocessing data in a traffic flow data set, and constructing flow sub-tensors under attribute features and three time granularities; then, for each region, a region association graph is constructed, including a distance graph and a semantic graph. And finally, constructing a space-time network block, obtaining space-time representations under the time granularity through the space-time network block on the basis of the distance map and the semantic map in combination with the attribute features and the traffic sub-tensors under the three time granularities, fusing the space-time representations under the time granularities to predict the traffic, and performing reverse training. According to the method, the multi-modal spatial association relationship is better captured, and the accuracy of urban area traffic flow prediction is improved.

    本发明公开了一种考虑多时空粒度的城市区域交通流量预测方法,该方法首先获取交通流量数据集、区域数据集、天气数据集和兴趣点POI数据集。其次对交通流量数据集中数据进行预处理,并构建属性特征和三个时间粒度下的流量子张量。然后对于每个区域,构建区域关联图,包括距离图和语义图。最后构建时空网络块,基于距离图和语义图,结合属性特征和三个时间粒度下的流量子张量,通过时空网络块,得到时间粒度下的时空表示,融合各时间粒度下的时空表示预测流量,并进行反向训练。本发明更好地捕获多模态的空间关联关系,提升城市区域交通流量预测的准确率。


    Access

    Download


    Export, share and cite



    Title :

    Urban area traffic flow prediction method considering multiple space-time granularities


    Additional title:

    一种考虑多时空粒度的城市区域交通流量预测方法


    Contributors:
    SHI XIAOYING (author) / LI GUOYI (author) / HE BOWEN (author) / XU HAITAO (author) / TU LIMING (author) / LIU GENG (author)

    Publication date :

    2024-12-20


    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