Augmented Multi-Component Recurrent Graph Convolutional Network for Traffic Flow Forecasting

doi:10.3390/ijgi11020088

	Augmented Multi-Component Recurrent Graph Convolutional Network for Traffic Flow Forecasting
	Zhang, Chi 1; Zhou, Hong-Yu 2; Qiu, Qiang 3; Jian, Zhichun 4; Zhu, Daoye 1; Cheng, Chengqi 5; He, Liesong 6; Liu, Guoping 7; Wen, Xiang 7; Hu, Runbo 7
	2022-02-01
发表期刊	ISPRS INTERNATIONAL JOURNAL OF GEO-INFORMATION
卷号	11 期号:2 页码:17
摘要	Due to the periodic and dynamic changes of traffic flow and the spatial-temporal coupling interaction of complex road networks, traffic flow forecasting is highly challenging and rarely yields satisfactory prediction results. In this paper, we propose a novel methodology named the Augmented Multi-component Recurrent Graph Convolutional Network (AM-RGCN) for traffic flow forecasting by addressing the problems above. We first introduce the augmented multi-component module to the traffic forecasting model to tackle the problem of periodic temporal shift emerging in traffic series. Then, we propose an encoder-decoder architecture for spatial-temporal prediction. Specifically, we propose the Temporal Correlation Learner (TCL) which incorporates one-dimensional convolution into LSTM to utilize the intrinsic temporal characteristics of traffic flow. Moreover, we combine TCL with the graph convolutional network to handle the spatial-temporal coupling interaction of the road network. Similarly, the decoder consists of TCL and convolutional neural networks to obtain high-dimensional representations from multi-step predictions based on spatial-temporal sequences. Extensive experiments on two real-world road traffic datasets, PEMSD4 and PEMSD8, demonstrate that our AM-RGCN achieves the best results.
关键词	traffic flow forecasting spatial-temporal prediction graph convolutional networks augmented multi-component
DOI	10.3390/ijgi11020088
收录类别	SCI
语种	英语
资助项目	Science and Technology Major Special Project of Guangxi, China[GUIKEAA18118025]
WOS研究方向	Computer Science ; Physical Geography ; Remote Sensing
WOS类目	Computer Science, Information Systems ; Geography, Physical ; Remote Sensing
WOS记录号	WOS:000762474900001
出版者	MDPI
引用统计	被引频次：6[WOS] [WOS记录] [WOS相关记录]
文献类型	期刊论文
条目标识符	http://119.78.100.204/handle/2XEOYT63/18915
专题	中国科学院计算技术研究所期刊论文_英文
通讯作者	Jian, Zhichun
作者单位	1.Peking Univ, Acad Adv Interdisciplinary Studies, Beijing 100871, Peoples R China 2.Univ Hong Kong, Dept Comp Sci, Hong Kong 999077, Peoples R China 3.Chinese Acad Sci, Inst Comp Technol, Beijing 101408, Peoples R China 4.Shopee Informat Technol Co Ltd, Shenzhen 518063, Peoples R China 5.Peking Univ, Coll Engn, Beijing 100871, Peoples R China 6.Xian Res Inst Surveying & Mapping, Xian 710000, Peoples R China 7.Didi Chuxing, Beijing 100085, Peoples R China
推荐引用方式 GB/T 7714	Zhang, Chi,Zhou, Hong-Yu,Qiu, Qiang,et al. Augmented Multi-Component Recurrent Graph Convolutional Network for Traffic Flow Forecasting[J]. ISPRS INTERNATIONAL JOURNAL OF GEO-INFORMATION,2022,11(2):17.
APA	Zhang, Chi.,Zhou, Hong-Yu.,Qiu, Qiang.,Jian, Zhichun.,Zhu, Daoye.,...&Hu, Runbo.(2022).Augmented Multi-Component Recurrent Graph Convolutional Network for Traffic Flow Forecasting.ISPRS INTERNATIONAL JOURNAL OF GEO-INFORMATION,11(2),17.
MLA	Zhang, Chi,et al."Augmented Multi-Component Recurrent Graph Convolutional Network for Traffic Flow Forecasting".ISPRS INTERNATIONAL JOURNAL OF GEO-INFORMATION 11.2(2022):17.