摘要
文中提出了一种基于节点向量-生成对抗网络的交通流预测方法.通过Node2vec方法实现路网邻接关系的重构,实现路网空间相关性的深度挖掘.基于残差图聚合机制构建了路网数据空间特征的生成器,实现了根据路网中的部分已知数据推演未来路网交通流数据.采用西雅图高速路网速度数据集(Seattle)和加州路网速度数据集(PEMS)验证模型的有效性.结果表明:该模型在不同数据缺失模式、不同数据缺失率下均可以保持鲁棒的交通流预测表现.
A traffic flow forecasting method based on node vector-generation countermeasure network was proposed.The reconstruction of road network adjacency relationship was realized by Node2vec method,and the deep mining of road network spatial correlation was realized.Based on the residual graph aggregation mechanism,a generator of spatial characteristics of road network data was constructed,and the future road network traffic flow data was deduced according to some known data in the road network.Seattle Expressway Network Speed Data Set(Seattle)and California Highway Network Speed Data Set(PEMS)were used to verify the effectiveness of the model.The results show that the model can maintain robust traffic flow prediction performance under different data missing modes and different data missing rates.
作者
徐东伟
朱宏俊
周磊
杨艳芳
XU Dongwei;ZHU Hongjun;ZHOU Lei;YANG Fangyan(Research Center of Urban Traffic Information Intelligent Sensing and Service Technology,Beijing 100084,China;Institute of Cyberspace Security,Zhejiang University of Technology,Hangzhou 310000,China;College of Information Engineering,Zhejiang University of Technology,Hangzhou 310000,China;Scientific Research Institute of the Ministry of Transport,Beijing 100029,China)
出处
《武汉理工大学学报(交通科学与工程版)》
2024年第2期211-217,共7页
Journal of Wuhan University of Technology(Transportation Science & Engineering)
基金
国家自然科学基金青年科学基金(61903334)
浙江省自然科学基金(LY21F030016,LY16F030016)
综合交通运输大数据应用技术交通运输行业重点实验室开放课题基金(2020B1205)
北京市城市交通信息智能感知与服务工程技术研究中心开放课题基金(UTIS2023KF01)
中央级公益性科研院所基本科研业务费项目(20221204)。
关键词
智能交通
交通流预测
Node2vec
数据缺失
生成对抗网络
intelligent transportation system
traffic flow prediction
Node2vec
missing data
generative adversarial nets