城市热点站点交通流量预测模型研究

Research on traffic flow prediction model of urban hot spots

在城市局部及热点区域,交通预测受站点自身特征信息影响显著。针对上述问题,文章针对城市站点应用环境提出了一种交通流量预测模型,能够结合站点自身特征信息完成特征分析,其中空间、时间特征分别采用适应图卷积(daptive graph convolution,DGCN)模块、门控扩张因果卷积进行提取,增大层深度来获得更大感受野,进而得到更为丰富的节点时间特征。对比实验、消融实验结果表明,CS-DSTGCN预测模型在城市站点环境下的交通流量预测具有突出应用价值及优势。

In the local and hot spots of the city,traffic prediction is significantly affected by the characteristics of the site itself.In view of the above problems,this paper proposes a traffic flow prediction model for the application environment of urban sites,which can combine the characteristics of the site itself to complete the feature analysis.The spatial and temporal features are extracted by the adaptive graph convolution(DGCN)module and the gated expansion causal convolution respectively,and the layer depth is increased to obtain a larger receptive field,and then a richer node time feature is obtained.The results of comparative experiments and ablation experiments show that the CS-DSTGCN prediction model has outstanding application value and advantages in traffic flow prediction in urban site environment.