基于时序分解和注意力图神经网络的交通预测  

Time-Series Decomposition and Attention Graph Neural Network Based Traffic Forecasting

作　　者：杨永鹏 杨震[1] 杨真真[1,2] YANG Yongpeng;YANG Zhen;YANG Zhenzhen(School of Communications and Information Engineering,Nanjing University of Posts and Telecommunications,Nanjing 210023,China;College of Science,Nanjing University of Posts and Telecommunications,Nanjing 210023,China)

机构地区：[1]南京邮电大学通信与信息工程学院,南京210023 [2]南京邮电大学理学院,南京210023

出　　处：《数据采集与处理》2025年第2期417-430,共14页Journal of Data Acquisition and Processing

基　　金：国家自然科学基金(62071242,62171232);江苏省研究生科研与实践创新计划项目(KYCX22_0955)。

摘　　要：如何有效挖掘隐藏在交通数据中的时空依赖信息、动态信息和空间异质信息一直是交通预测任务面临的关键问题。本文提出了一种基于时序分解和注意力图神经网络(Time‑series decomposition and attention graph neural network,TDAGNN)的交通预测模型。采用双分支时序分解卷积神经网络(Dual time‑series decomposition convolutional neural network,DTDCNN)从复杂的交通数据中挖掘时间依赖信息;采用多头交互注意力网络(Multi‑head interactive attention,MIA)对原始交通特征和局部增强特征进行交互学习,以深入挖掘交通数据的异质信息和动态信息;引入自缩放动态扩散图神经网络(Self‑scaling dynamic diffusion graph neural network,SDDGNN)在获取交通数据空间依赖信息的同时,避免图神经网络的尺度失真问题;将提出的TDAGNN应用于经典交通数据PEMS04、PEMS08、METR‑LA和PEMS‑BAY的交通预测实验中。实验结果表明,提出模型的平均MAE、RMSE和MAPE比其他经典算法最大可分别提高14.64、23.68和9.41%,从而证明其具有较高的交通预测精度。In order to address challenges on how to accurately capture the spatial‑temporal dependency,dynamic information and spatial heterogeneity information in traffic forecasting,we propose a time‑series decomposition and attention graph neural network(TDAGNN)based traffic forecasting.Specifically,the model first adopts the dual time‑series decomposition convolutional neural network(DTDCNN)to extract temporal dependency from traffic data.Secondly,the multi‑head interactive attention(MIA)network is introduced to capture spatial heterogeneity and dynamicity from traffic data via the interactivity between the original features and the local augmentation features.Thirdly,the self‑scaling dynamic diffusion graph neural network(SDDGNN)is introduced for capturing the spatial dependence and dynamicity from the traffic data.Finally,extensive experiments are carried out for some datasets.Experimental results demonstrate that the average MAE,RMSE and MAPE of the proposed model can be improved up to 14.64,23.68 and 9.41%respectively,compared to other classic algorithms,proving its high prediction accuracy.

关 键 词：交通预测 时序分解 图神经网络 注意力机制 局部增强网络 

分 类 号：TP391[自动化与计算机技术—计算机应用技术] TP183[自动化与计算机技术—计算机科学与技术]