图神经网络驱动的交通预测技术：探索与挑战

doi:10.11959/j.issn.2096-3750.2021.00235

物联网学报 ›› 2021, Vol. 5 ›› Issue (4): 1-16.doi: 10.11959/j.issn.2096-3750.2021.00235

• 前沿与综述 • 下一篇

图神经网络驱动的交通预测技术：探索与挑战

周毅¹, 胡姝婷¹, 李伟¹, 承楠², 路宁³, 沈学民⁴

¹ 河南大学，河南郑州 450046
² 西安电子科技大学，陕西西安 710071
³ 加拿大女王大学，加拿大金斯顿 K7L 3N6
⁴ 滑铁卢大学，加拿大滑铁卢 N2L 3G1

修回日期:2021-07-24 出版日期:2021-12-30 发布日期:2021-12-01
作者简介:周毅（1981− ），男，博士，河南大学教授、博士生导师，河南省车联网协同技术国际联合实验室主任，主要研究方向为车联网与智能交通、空地协同组网、平行增强学习、协作机器人等
胡姝婷（1997− ），女，河南大学硕士生，主要研究方向为图神经网络、智能交通等
李伟（1979− ），女，河南大学副教授、硕士生导师，主要研究方向为车联网优化控制、协作通信等
承楠（1987− ），男，博士，西安电子科技大学教授、博士生导师，主要研究方向为车联网与先进交通系统、人工智能、空天地一体化网络等
路宁（1984− ），男，博士，加拿大女王大学助理教授，主要研究方向为车联网与智能交通、深度强化学习、移动边缘计算等
沈学民（1958− ），男，博士，中国工程院外籍院士，加拿大工程院、工程研究院院士以及皇家学会会士，加拿大滑铁卢大学教授，IEEE Fellow，主要研究方向为空天地一体化网络、车联网、网络安全、人工智能及智能电网等
基金资助:
国家自然科学基金资助项目(62176088);国家自然科学基金资助项目(61701170);国家自然科学基金资助项目(62071356);河南省科技攻关项目(202102310198);河南省科技攻关项目(212102210412)

Graph neural network driven traffic prediction technology:review and challenge

Yi ZHOU¹, Shuting HU¹, Wei LI¹, Nan CHENG², Ning LU³, Xuemin(Sherman) SHEN⁴

¹ Henan University, Zhenzhou 450046, China
² Xidian University, Xi’an 710071, China
³ Queen’s University, Kingston K7L 3N6, Canada
⁴ University of Waterloo, Waterloo N2L 3G1, Canada

Revised:2021-07-24 Online:2021-12-30 Published:2021-12-01
Supported by:
The National Natural Science Foundation of China(62176088);The National Natural Science Foundation of China(61701170);The National Natural Science Foundation of China(62071356);The Henan Science and Technology Development Program(202102310198);The Henan Science and Technology Development Program(212102210412)

摘要/Abstract

摘要：

随着物联网及人工智能技术的快速发展，对交通数据进行精准的分析和预测成为智慧交通的首要环节。近年来，交通预测方法逐渐从经典的模型驱动转变为数据驱动，然而，如何通过大数据有效分析路网的时空特性是预测过程中面临的关键难题之一。时空大数据分析是交通预测的利器，将交通路网建模为图网络，将深度学习方法在图网络上进行扩展，通过图神经网络建立时空预测模型，采用图卷积的方式有效地获取路网传感器节点之间的时空相关性，可以显著提高交通预测模型的精度。针对图神经网络驱动的交通预测技术进行了探索，基于深度时空特性分析提炼了两大类交通预测模型，并通过实例进行分析和验证，探讨了图神经网络在交通预测领域的技术优势和主要挑战，挖掘了图神经网络预测机制的潜在研究方向。

关键词: 交通预测, 图神经网络, 时空相关性, 同步卷积, 图注意力网络

Abstract:

With the rapid development of Internet of things and artificial intelligence technology, accurate analysis and prediction of traffic data have become the primary target of intelligent transportations.In recent years, the method of traffic forecasting has gradually changed from the classical model-driven type to the data-driven type.However, how to effectively analyze the spatial-temporal characteristics of road networks through big data is one of the key issues in the traffic prediction process.Spatiotemporal big data analysis is a powerful tool for the traffic prediction.The traffic network can be modeled as a graph network, while the deep learning method can be extended on the graph network.Utilizing graph neural networks, we can build the spatiotemporal prediction model, and obtain the spatial-temporal correlation between the sensor nodes in road networks effectively by using graph convolution, which can significantly improve the accuracy of traffic prediction models.The traffic forecasting technology driven by graph neural networks was explored, and two kinds of traffic prediction models based on the analysis of deep spatial-temporal characteristics were extracted.The actual cases were analyzed and evaluated to discuss the technical advantages and key challenges of graph neural networks in the traffic prediction.The potential issues of graph neural network driven prediction mechanisms were also excavated.

Key words: traffic prediction, graph neural networks, spatial-temporal correlation, synchronous convolution, graph at-tention networks

中图分类号:

TP391

周毅, 胡姝婷, 李伟, 承楠, 路宁, 沈学民. 图神经网络驱动的交通预测技术：探索与挑战[J]. 物联网学报, 2021, 5(4): 1-16.

Yi ZHOU, Shuting HU, Wei LI, Nan CHENG, Ning LU, Xuemin(Sherman) SHEN. Graph neural network driven traffic prediction technology:review and challenge[J]. Chinese Journal on Internet of Things, 2021, 5(4): 1-16.

图/表 15

表1

表2

图1

图2

图3

图4

图5

图6

图7

图8

图9

图10

图11

图12

表3

不同预测方法的性能比较"

时间	模型	MAE	RMSE	MAPE
15 min	HA	2.88	5.59	6.76%
	VAR	1.74	3.09	3.59%
	ARIMA	1.62	3.30	3.50%
	FC-LSTM	2.05	4.19	4.80%
	DCRNN	1.31	2.76	2.73%
	Graph WaveNet	$1 . 30$	$2 . 74$	$2 . 73 %$
30 min	HA	2.88	5.59	6.76%
	VAR	2.33	4.15	5.02%
	ARIMA	2.33	4.76	5.40%
	FC-LSTM	2.20	4.55	5.20%
	DCRNN	1.65	3.77	3.72%
	Graph WaveNet	$1 . 63$	$3 . 70$	$3 . 67 %$
60 min	HA	2.88	5.59	6.76%
	VAR	2.92	5.11	6.46%
	ARIMA	3.38	6.50	8.30%
	FC-LSTM	2.37	4.96	5.70%
	DCRNN	1.97	4.62	4.71%
	Graph WaveNet	$1 . 95$	$4 . 52$	$4 . 63 %$

表3

参考文献 69

[1]	LU N , CHENG N , ZHANG N ,et al. Connected vehicles:solutions and challenges[J]. IEEE Internet of Things Journal, 2014,1(4): 289-299.
[2]	QI Y L , TIAN L , ZHOU Y Q ,et al. Mobile edge computing-assisted admission control in vehicular networks:the convergence of communication and computation[J]. IEEE Vehicular Technology Magazine, 2019,14(1): 37-44.
[3]	ZHOU Y Q , TIAN L , LIU L ,et al. Fog computing enabled future mobile communication networks:a convergence of communication and computing[J]. IEEE Communications Magazine, 2019,57(5): 20-27.
[4]	张士兵, 王婷婷, 张晓格 ,等. 智能交通车载网的现状及其发展策略[J]. 通信技术, 2017,50(7): 1345-1350.
	ZHANG S B , WANG T T , ZHANG X G ,et al. Development strategy and status of intelligent transportation vehicular networks[J]. Communications Technology, 2017,50(7): 1345-1350.
[5]	ZHAO L , SONG Y J , ZHANG C ,et al. T-GCN:a temporal graph convolutional network for traffic prediction[J]. IEEE Transactions on Intelligent Transportation Systems, 2020,21(9): 3848-3858.
[6]	余辰, 张丽娟, 金海 . 大数据驱动的智能交通系统研究进展与趋势[J]. 物联网学报, 2018,2(1): 56-63.
	YU C , ZHANG L J , JIN H . Research progress and trend of big da ta-driven intelligent transportation system[J]. Chinese Journal on Internet of Things, 2018,2(1): 56-63.
[7]	殷礼胜, 唐圣期, 李胜 ,等. 基于整合移动平均自回归和遗传粒子群优化小波神经网络组合模型的交通流预测[J]. 电子与信息学报, 2019,41(9): 2273-2279.
	YIN L S , TANG S Q ， LISHENG ,et al. Traffic flow prediction based on hybrid model of auto-regressive integrated moving average and genetic particle swarm optimization wavelet neural network[J]. Journal of Electronics ＆ Information Technology, 2019,41(9): 2273-2279.
[8]	GUO J H , HUANG W , WILLIAMS B M . Adaptive Kalman filter approach for stochastic short-term traffic flow rate prediction and uncertainty quantification[J]. Transportation Research Part C:Emerging Technologies, 2014,43: 50-64.
[9]	DELL'ACQUA P , BELLOTTI F , BERTA R ,et al. Time-aware multivariate nearest neighbor regression methods for traffic flow prediction[J]. IEEE Transactions on Intelligent Transportation Systems, 2015,16(6): 3393-3402.
[10]	LUO X L , LI D Y , ZHANG S R . Traffic flow prediction during the holidays based on DFT and SVR[J]. Journal of Sensors, 2019,2019: 1-10.
[11]	DUAN Y J , L V Y , WANG F Y . Travel time prediction with LSTM neural network[C]// Proceedings of 2016 IEEE 19th International Conference on Intelligent Transportation Systems (ITSC). Piscataway:IEEE Press, 2016: 1053-1058.
[12]	WANG W , ZHOU C H , HE H L ,et al. Cellular traffic load prediction with LSTM and Gaussian process regression[C]// Proceedings of ICC 2020 - 2020 IEEE International Conference on Communications (ICC). Piscataway:IEEE Press, 2020: 1-6.
[13]	HAN L Y , ZHENG K , ZHAO L ,et al. Short-term traffic prediction based on DeepCluster in large-scale road networks[J]. IEEE Transactions on Vehicular Technology, 2019,68(12): 12301-12313.
[14]	SCARSELLI F , GORI M , TSOI A C ,et al. The graph neural network model[J]. IEEE Transactions on Neural Networks, 2009,20(1): 61-80.
[15]	WU Z H , PAN S R , CHEN F W ,et al. A comprehensive survey on graph neural networks[J]. IEEE Transactions on Neural Networks and Learning Systems, 2021,32(1): 4-24.
[16]	ZHOU J , CUI G Q , ZHANG Z Y ,et al. Graph neural networks:a review of methods and applications[J]. ArXiv:1812.08434, 2018.
[17]	ZHANG J W . Graph neural networks for small graph and giant network representation learning:an overview[J]. ArXiv:1908.00187, 2019.
[18]	KIPF T N , WELLING M . Semi-supervised classification with graph convolutional networks[C]// Proceedings of International Conference on Learning Representations (ICLR).[S.l.:s.n.], 2016.
[19]	LI Y G , YU R , SHAHABI C ,et al. Graph convolutional recurrent neural network:data-driven traffic forecasting[EB]. Proceedings of International Conference on Learning Representations (ICLR).[S.l.:s.n.], 2017.
[20]	YU B , YIN H T , ZHU Z X . Spatio-temporal graph convolutional networks:a deep learning framework for traffic forecasting[C]// Proceedings of the Twenty-Seventh International Joint Conference on Artificial Intelligence. California:International Joint Conferences on Artificial Intelligence Organization, 2018.
[21]	GUO K , HU Y L , QIAN Z ,et al. Optimized graph convolution recurrent neural network for traffic prediction[J]. IEEE Transactions on Intelligent Transportation Systems, 2021,22(2): 1138-1149.
[22]	CIRSTEA R G , GUO C J , YANG B ,et al. Graph attention recurrent neural networks for correlated time series forecasting[EB]. 2021.
[23]	冯宁, 郭晟楠, 宋超 ,等. 面向交通流量预测的多组件时空图卷积网络[J]. 软件学报, 2019,30(3): 759-769.
	FENG N , GUO S N , SONG C ,et al. Multi-component spa tial-temporal graph convolution networks for traffic flow forecasting[J]. Journal of Software, 2019,30(3): 759-769.
[24]	GUO S N , LIN Y F , FENG N ,et al. Attention based spatial-temporal graph convolutional networks for traffic flow forecasting[J]. Proceedings of the AAAI Conference on Artificial Intelligence, 2019,33: 922-929.
[25]	ZHOU F , YANG Q , ZHANG K P ,et al. Reinforced spatiotemporal attentive graph neural networks for traffic forecasting[J]. IEEE Internet of Things Journal, 2020,7(7): 6414-6428.
[26]	SHI X M , QI H , SHEN Y M ,et al. A spatial-temporal attention approach for traffic prediction[J]. IEEE Transactions on Intelligent Transportation Systems, 2020.
[27]	ZHAO J D , GAO Y , YANG Z Z ,et al. Truck traffic speed prediction under non-recurrent congestion:based on optimized deep learning algorithms and GPS data[J]. IEEE Access, 2019,7: 9116-9127.
[28]	GUO J L , SONG C Y , WANG H . A multi-step traffic speed forecasting model based on graph convolutional LSTM[C]// Proceedings of 2019 Chinese Automation Congress (CAC). Piscataway:IEEE Press, 2019: 2466-2471.
[29]	GE L , LI H , LIU J L ,et al. Temporal graph convolutional networks for traffic speed prediction considering external factors[C]// Proceedings of 2019 20th IEEE International Conference on Mobile Data Management (MDM). Piscataway:IEEE Press, 2019: 234-242.
[30]	WU Z H , PAN S R , LONG G D ,et al. Graph WaveNet for deep spatial-temporal graph modeling[C]// Proceedings of the Twenty-Eighth International Joint Conference on Artificial Intelligence. California:International Joint Conferences on Artificial Intelligence Organization, 2019: 1907-1913.
[31]	CUI Z Y , HENRICKSON K , KE R M ,et al. Traffic graph convolutional recurrent neural network:a deep learning framework for network-scale traffic learning and forecasting[J]. IEEE Transactions on Intelligent Transportation Systems, 2020,21(11): 4883-4894.
[32]	ZHANG Z C , LI M , LIN X ,et al. Multistep speed prediction on traffic networks:a deep learning approach considering spatio-temporal dependencies[J]. Transportation Research Part C:Emerging Technologies, 2019,105: 297-322.
[33]	ZHANG C H , YU J J Q , LIU Y . Spatial-temporal graph attention networks:a deep learning approach for traffic forecasting[J]. IEEE Access, 2019,7: 166246-166256.
[34]	ZHENG C P , FAN X L , WANG C ,et al. GMAN:a graph multi-attention network for traffic prediction[C]// Proceedings of the AAAI Conference on Artificial Intelligence.[S.l.:s.n.], 2020,34(1): 1234-1241.
[35]	SONG C , LIN Y F , GUO S N ,et al. Spatial-temporal synchronous graph convolutional networks:a new framework for spatial-temporal network data forecasting[C]// Proceedings of the AAAI Conference on Artificial Intelligence.[S.l.:s.n.], 2020,34(1): 914-921.
[36]	Park C , Lee C , Bahng H ,et al. STGRAT:a spatio-temporal graph attention network for traffic forecasting[J]. ArXiv:1911.13181, 2019.
[37]	SUTSKEVER I , VINYALS O , LE Q V . Sequence to sequence learning with neural networks[C]// Proceedings of Advances in Neural Information Processing Systems.[S.l.:s.n.], 2014: 3104-3112.
[38]	BENGIO S , VINYALS O , JAITLY N ,et al. Scheduled sampling for sequence prediction with recurrent neural networks[C]// Proceedings of Advances in Neural Information Processing Systems.[S.l.:s.n.], 2015: 1171-1179.
[39]	RANZATO M , CHOPRA S , AULI M ,et al. Sequence level training with recurrent neural networks[J]. Arxiv:1511.06732, 2015.
[40]	DAUMé H , LANGFORD J , MARCU D . Search-based structured prediction[J]. Machine Learning, 2009,75(3): 297-325.
[41]	YU L T , ZHANG W N , WANG J ,et al. SeqGAN:sequence generative adversarial nets with policy gradient[C]// Proceedings of 31st AAAI Conference on Artificial Intelligence.[S.l.:s.n.], 2016.
[42]	KENESHLOO Y , SHI T , RAMAKRISHNAN N ,et al. Deep reinforcement learning for sequence-to-sequence models[J]. IEEE Transactions on Neural Networks and Learning Systems, 2020,31(7): 2469-2489.
[43]	LANA I , DEL SER J , VELEZ M ,et al. Road traffic forecasting:recent advances and new challenges[J]. IEEE Intelligent Transportation Systems Magazine, 2018,10(2): 93-109.
[44]	NAGY A M , SIMON V . Survey on traffic prediction in smart cities[J]. Pervasive and Mobile Computing, 2018,50: 148-163.
[45]	ZHANG Y J . Short-term traffic flow prediction methods:a survey[J]. Journal of Physics:Conference Series,2020, 1486:052018.
[46]	XIE P , LI T R , LIU J ,et al. Urban flow prediction from spatiotemporal data using machine learning:a survey[J]. Information Fusion, 2020,59: 1-12.
[47]	YE J , ZHAO J , YE K ,et al. How to build a graph-based deep learning architecture in traffic domain:a survey[J]. IEEE Transactions on Intelligent Transportation Systems, 2020.
[48]	ZHANG Z , CUI P , ZHU W . Deep learning on graphs:a survey[J]. IEEE Transactions on Knowledge and Data Engineering, 2020.
[49]	JAIN A , ZAMIR A R , SAVARESE S ,et al. Structural-RNN:deep learning on spatio-temporal graphs[C]// Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Piscataway:IEEE Press, 2016: 5308-5317.
[50]	HOCHREITER S , SCHMIDHUBER J . Long short-term memory[J]. Neural Computation, 1997,9(8): 1735-1780.
[51]	王祥雪, 许伦辉 . 基于深度学习的短时交通流预测研究[J]. 交通运输系统工程与信息, 2018,18(1): 81-88.
	WANG X X , XU L H . Short-term traffic flow prediction based on deep learning[J]. Journal of Transportation Systems Engineering and Information Technology, 2018,18(1): 81-88.
[52]	MA X L , TAO Z M , WANG Y H ,et al. Long short-term memory neural network for traffic speed prediction using remote microwave sensor data[J]. Transportation Research Part C:Emerging Technologies, 2015,54: 187-197.
[53]	TIAN Y X , PAN L . Predicting short-term traffic flow by long short-term memory recurrent neural network[C]// Proceedings of 2015 IEEE International Conference on Smart City/SocialCom/SustainCom (SmartCity). Piscataway:IEEE Press, 2015: 153-158.
[54]	YU R , LI Y G , SHAHABI C ,et al. Deep learning:A generic approach for extreme condition traffic forecasting[C]// Proceedings of the 2017 SIAM International Conference on Data Mining. Philadelphia,PA:Society for Industrial and Applied Mathematics, 2017: 777-785.
[55]	FU R , ZHANG Z , LI L . Using LSTM and GRU neural network methods for traffic flow prediction[C]// Proceedings of 2016 31st Youth Academic Annual Conference of Chinese Association of Automation (YAC). Piscataway:IEEE Press, 2016: 324-328.
[56]	LECUN Y , BENGIO Y . Convolutional networks for images,speech,and time-series[J]. The Handbook of Brain Theory and Neural Networks, 1995,3361(10): 1-14.
[57]	DEFFERRARD M , BRESSON X , VANDERGHEYNST P . Convolutional neural networks on graphs with fast localized spectral filtering[J]. Advances in Neural Information Processing Systems, 2016,29: 3844-3852.
[58]	VELI?KOVI? P , CUCURULL G , CASANOVA A ,et al. Graph attention networks[J]. Arxiv:1710.10903, 2017.
[59]	BUSBRIDGE D , SHERBURN D , CAVALLO P ,et al. Relational graph attention networks[J]. Arxiv:1904.05811, 2019.
[60]	VASWANI A , SHAZEER N , PARMAR N ,et al. Attention is all you need[J]. Advances in Neural Information Processing Systems, 2017,30: 5998-6008.
[61]	BELLO I , ZOPH B , LE Q ,et al. Attention augmented convolutional networks[C]// Proceedings of 2019 IEEE/CVF International Conference on Computer Vision. Piscataway:IEEE Press, 2019: 3285-3294.
[62]	CHEN C , VARAIYA P . Freeway performance measurement system (PeMS)[EB]. 2003.
[63]	WU W , CHEN N , ZHOU C H ,et al. Dynamic RAN slicing for service-oriented vehicular networks via constrained learning[J]. IEEE Journal on Selected Areas in Communications, 2021,39(7): 2076-2089.
[64]	BAHDANAU D , BRAKEL P , XU K ,et al. An actor-critic algorithm for sequence prediction[J]. arXiv:1607.07086, 2016.
[65]	MALLICK T , BALAPRAKASH P , RASK E ,et al. Graph- partitioning-based diffusion convolutional recurrent neural network for large-scale traffic forecasting[J]. Transportation Research Record:Journal of the Transportation Research Board, 2020,2674(9): 473-488.
[66]	BAI S , KOLTER J Z , KOLTUN V . An empirical evaluation of generic convolutional and recurrent networks for sequence modeling[J]. Arxiv:1803.01271, 2018.
[67]	LEA C , FLYNN M D , VIDAL R ,et al. Temporal convolutional networks for action segmentation and detection[C]// Proceedings of 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway:IEEE Press, 2017: 1003-1012.
[68]	王坤峰, 苟超, 段艳杰 ,等. 生成式对抗网络GAN的研究进展与展望[J]. 自动化学报, 2017,43(3): 321-332.
	WANG K F , GOU C , DUAN Y J ,et al. Generative adversarial net works:the state of the art and beyond[J]. Acta Automatica Sinica, 2017,43(3): 321-332.
[69]	ZHOU Y Q , LIU L , WANG L ,et al. Service-aware 6G:an intelligent and open network based on the convergence of communication,computing and caching[J]. Digital Communications and Networks, 2020,6(3): 253-260.

方法		优点	缺点
模型驱动		主要研究流量、速度和密度之间的瞬时和稳态关系，主要依赖先验知识进行系统建模	模型固定，难以准确地模拟变化多端的真实交通状况
数据驱动	统计分析模型	算法和模型简易，实现方便	模型建立在时间序列数据平稳的假设前提下
	机器学习模型	可以提取与交通相关的特征，用于非线性数据的模型构建	依赖于人工提取的交通特征，模型架构浅易、参数有限且计算效率较低
	深度学习模型	能够让计算机自动学习路网特征，减少对人工提取特征的依赖性	只考虑路网空间或时间特征
	图神经网络+深度学习	从时间和空间两个角度提取路网特征	学习时间相关性效率不高，长期预测精度有待进一步提高

图神经网络驱动的交通预测技术：探索与挑战

Graph neural network driven traffic prediction technology:review and challenge

在线阅读

PDF下载

可视化

摘要/Abstract

引用本文

使用本文

图/表 15

参考文献 69

相关文章 1

Metrics

推荐阅读 0

文献	空间模型	时间模型	数据集
[5]	GCN	GRU	SZ-taxi；Los-loop
[28]	GCN	LSTM	出租车的GPS数据（上海TIC）
[23]	GCN	标准二维卷积	PeMSD4；PeMSD8
[20]	GCN	门控CNN	BJER4；PeMSD7
[29]	GCN	DCC	PeMSD4；PeMSD7
[30]	GCN	门控TCN	METR-LA；PEMS-BAY
[31]	考虑高阶邻域的GCN	LSTM	LOOP；INRIX
[32]	GCN	GRU+注意力	A-map
[19]	扩散卷积网络	GRU	METR-LA；PEMS-BAY
[21]	基于优化图矩阵的GCN	GRU	D.C.；Philadelphia；PeMSD4
[22]	GAT	GRU	METR-LA
[24]	GCN+注意力	标准二维卷积+注意力	PeMSD4；PeMSD8
[26]	注意力	LSTM+注意力	PeMSD4；PeMSD8
[33]	GAT	LSTM	PeMSD7
[34]	注意力	注意力	Xiamen；PeMS
[35]	时空同步卷积	时空同步卷积	PEMS03；PEMS04；PEMS07；PEMS08
[36]	注意力	注意力	METR-LA；PEMS-BAY