一种用于交通预测的注意力时空图神经网络

doi:10.19665/j.issn1001-2400.2023.01.019

西安电子科技大学学报 ›› 2023, Vol. 50 ›› Issue (1): 168-176.doi: 10.19665/j.issn1001-2400.2023.01.019

一种用于交通预测的注意力时空图神经网络

甘萍¹(),农丽萍^2,³(),张文辉⁴(),林基明^1,²(),王俊义¹()

1.桂林电子科技大学信息与通信学院,广西壮族自治区桂林 541004
2.西安电子科技大学通信工程学院,陕西西安 710071
3.广西师范大学物理科学与技术学院,广西壮族自治区桂林 541001
4.桂林电子科技大学计算机与信息安全学院,广西壮族自治区桂林 541004

收稿日期:2021-10-12 出版日期:2023-02-20 发布日期:2023-03-21
通讯作者: 张文辉(1970—),女,副教授,E-mail:zhangwh@guet.edu.cn
作者简介:甘萍(1997—),女,桂林电子科技大学硕士研究生,E-mail:ganping97@163.com;|农丽萍(1985—),女,讲师,E-mail:nongliping@mailbox.gxnu.edu.cn;|林基明(1970—),男,教授,E-mail:linjm@guet.edu.cn;|王俊义(1977—),男,研究员,E-mail:wangjy@guet.edu.cn
基金资助:
国家自然科学基金(61966007);广西自然科学基金(2020GXNSFAA159105);广西无线宽带通信与信号处理重点实验室基金(GXKL06190204);广西无线宽带通信与信号处理重点实验室基金(GXKL06200116);广西无线宽带通信与信号处理重点实验室基金(GXKL06190117);“认知无线电与信息处理”教育部重点实验室基金(CRKL180201);“认知无线电与信息处理”教育部重点实验室基金(CRKL180106)

Attention spatial-temporal graph neural network for traffic prediction

GAN Ping¹(),NONG Liping^2,³(),ZHANG Wenhui⁴(),LIN Jiming^1,²(),WANG Junyi¹()

1. School of Information and Communication,Guilin University of Electronic Technology,Guilin 541004,China
2. School of Telecommunications Engineering,Xidian University,Xi’an 710071,China
3. College of Physics and Technology,Guangxi Normal University,Guilin 541001,China
4. School of Computer Science and Information Security,Guilin University of Electronic Technology,Guilin 541004,China

Received:2021-10-12 Online:2023-02-20 Published:2023-03-21

摘要/Abstract

摘要：

城市化的发展使得交通预测在交通规划和城市管理等应用中发挥着重要作用。然而在交通预测任务中,捕获交通数据的高度非线性和复杂的时空依赖关系仍具有很大的挑战性。为了更好地捕获交通数据的时间依赖性和全局空间相关性以及同时满足长期和短期的预测任务,设计了一种用于交通预测的注意力时空图神经网络。首先通过引入注意力机制来调整邻近道路与非邻近道路的重要性,整合全局空间信息;然后再通过图卷积网络和带有扩展因果卷积的门控线性单元来共同捕获时空相关性。在两个真实数据集PeMSD7(M)和PEMS-BAY上的实验结果表明,该网络模型可以较大地提高短期和长期的交通预测精度。

关键词: 交通预测, 时空相关性, 注意力机制, 图卷积网络

Abstract:

With the development of urbanization,traffic prediction plays an important role in the application of traffic planning and urban management.However,in the task of traffic prediction,it is still a great challenge to capture the highly nonlinear and complex spatio-temporal dependencies of traffic data.In order to effectively capture the time dynamics and global spatial correlation of traffic data and satisfy both long-term and short-term prediction tasks,an attention based spatial-temporal graph neural network for traffic prediction is designed.First,the attention mechanism is introduced to adjust the importance of adjacent roads and non-adjacent roads,which is beneficial to integrating global spatial information.Then,the spatial-temporal correlations are captured by graph convolutional networks and gated linear units with extended causal convolution.Experimental results on two real data sets PeMSD7(M) and PEMS-BAY show that the network model can improve the accuracy of both long and short-term traffic prediction.

Key words: traffic prediction, spatial-temporal correlativity, mechanisms of attention, graph convolutional networks

中图分类号:

TP391

甘萍, 农丽萍, 张文辉, 林基明, 王俊义. 一种用于交通预测的注意力时空图神经网络[J]. 西安电子科技大学学报, 2023, 50(1): 168-176.

GAN Ping, NONG Liping, ZHANG Wenhui, LIN Jiming, WANG Junyi. Attention spatial-temporal graph neural network for traffic prediction[J]. Journal of Xidian University, 2023, 50(1): 168-176.

图/表 10

图1

图2

表1

表2

图3

图4

图5

图6

表3

表4

参考文献 19

[1]	HODGE V J, KRISHNAN R, AUSTIN J, et al. Short-Term Prediction of Traffic Flow Using a Binary Neural Network[J]. Neural Computing and Applications, 2014, 25(8):1639-1655. doi: 10.1007/s00521-014-1646-5
[2]	SUN H, ZHANG C, RAN B. Interval Prediction for Traffic Time Series Using Local Linear Predictor[C]// Proceedings of the 7th International IEEE Conference on Intelligent Transportation Systems.Piscataway:IEEE, 2004:410-415.
[3]	OKUTANI I, STEPHANEDES Y J. Dynamic Prediction of Traffic Volume Through Kalman Filtering Theory[J]. Transportation Research Part B:Methodological, 1984, 18(1):1-11. doi: 10.1016/0191-2615(84)90002-X
[4]	FU G, HAN G, LU F, et al. Short-Term Traffic Flow Forecasting Model Based on Support Vector Machine Regression[J]. Journal of South China University of Technology :Natural Science Edition, 2013, 41(9):71-76.
[5]	YIN H, WONG S C, XU J, et al. Urban Traffic Flow Prediction Using a Fuzzy-Neural Approach[J]. Transportation Research Part C:Emerging Technologies, 2002, 10(2):85-98. doi: 10.1016/S0968-090X(01)00004-3
[6]	SUN S, ZHANG C, YU G. A Bayesian Network Approach to Traffic Flow Forecasting[J]. IEEE Transactions on Intelligent Transportation Systems, 2006, 7(1):124-132. doi: 10.1109/TITS.2006.869623
[7]	NONG L P, WANG J Y, LIN J M, et al. Hypergraph Wavelet Neural Networks for 3D Object Classification[J]. Neurocomputing, 2021, 463:580-595. doi: 10.1016/j.neucom.2021.08.006
[8]	YU B, YIN H, ZHU Z. Spatio-Temporal Graph Convolutional Networks:A Deep Learning Framework for Traffic Forecasting[C]// Proceedings of the 27th International Joint Conference on Artificial Intelligence. San Francisco: Morgan Kaufmann, 2018:3634-3640.
[9]	LI Y, YU R, SHAHABI C, et al. Diffusion Convolutional Recurrent Neural Network:Data-Driven Traffic Forecasting[C/OL].[2018-2-22]. https://arxiv.org/abs/1707.01926v2.
[10]	ZHAO L, SONG Y, 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. doi: 10.1109/TITS.6979
[11]	GUO S, LIN Y, FENG N, et al. Attention Based Spatial-Temporal Graph Convolutional Networks for Traffic Flow Forecasting[C]// Proceedings of the AAAI Conference on Artificial Intelligence. Palo Alto: AAAI Press, 2019, 33(1):922-929.
[12]	SONG C, LIN Y, GUO S, 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. Palo Alto: AAAI Press, 2020:914-921.
[13]	VELIKOVI P, CUCURULL G, CASANOVA A, et al. Graph Attention Networks[C/OL].[2018-02-4]. https://arxiv.org/abs/1710.10903.
[14]	ZHU J, SONG Y, ZHAO L, et al. A3T-GCN:Attention Temporal Graph Convolutional Network for Traffic Forecasting[C/OL].[2020-06-20]. https://arxiv.org/abs/2006.11583.
[15]	WU Z, PAN S, LONG G, et al. Graph Wavenet for Deep Spatial-Temporal Graph Modeling[C]// Proceedings of the 28th International Joint Conference on Artificial Intelligence. San Francisco: Morgan Kaufmann, 2019:1907-1913.
[16]	LIU W, WU G, REN F, et al. DFF-ResNet:An Insect Pest Recognition Model Based on Residual Networks[J]. Big Data Mining and Analytics, 2020, 3(4):300-310. doi: 10.26599/BDMA.2020.9020021
[17]	HUANG R, HUANG C, LIU Y, et al. LSGCN:Long Short-Term Traffic Prediction with Graph Convolutional Networks[C]// Proceedings of the 29th International Joint Conference on Artificial Intelligence. San Francisco: Morgan Kaufmann. 2020:2355-2361.
[18]	XU M, DAI W, LIU C, et al. Spatial-Temporal Transformer Networks for Traffic Flow Forecasting[J/OL].[2020-01-14]. https://arxiv.org/abs/2001.02908.
[19]	SUTSKEVER I, VINYALS O, LE Q V. Sequence to Sequence Learning with Neural Networks[C]// Proceedings of the 27th International Conference on Neural Information Processing Systems. Cambridge: MIT Press, 2014:3104-3112.

数据集	#Nods	#Edges	#Time Steps
PeMSD7(M)	228	832	12 672
PEMS-BAY	325	2 369	52 116

模型	PeMSD7(M)(15/30/60 min)
模型	MAE	RMSE	MAPE/%
HA	4.01	7.20	10.61
ARIMA	5.55/5.86/6.75	9.00/9.13/9.72	12.92/13.94/16.92
FC-LSTM	3.57/3.94/4.35	6.20/7.03/7.93	8.60/9.55/10.61
STGCN	2.25/3.03/4.01	4.04/5.70/7.55	5.26/7.33/9.78
GraphWaveNet	2.14/2.80/3.42	4.01/5.48/6.72	4.93/6.89/8.57
LSGCN	2.22/2.96/3.81	3.98/5.47/7.09	5.14/7.18//9.60
STTN	2.14/2.70/3.31	4.04/5.37/6.51	5.05/6.68/7.98
AST-GCN	1.89/2.42/3.09	3.36/4.37/5.46	4.14/5.32/6.88
模型	PEMS-BAY(15/30/60 min)
模型	MAE	RMSE	MAPE/%
HA	2.88	5.59	6.80
ARIMA	1.62/2.23/3.38	3.30/4.76/6.50	3.50/5.40/8.30
FC-LSTM	2.05/2.20/2.37	4.19/4.55/4.96	4.80/5.20/5.70
STGCN	1.39/1.84/2.42	2.92/4.12/5.33	3.00/4.22/5.58
GraphWaveNet	1.30/1.63/1.95	2.73/3.67/4.63	2.74/3.70/4.52
STTN	1.36/1.67/1.95	2.87/3.79/4.50	2.89/3.78/4.58
AST-GCN	1.22/1.52/1.86	2.54/3.20/3.99	2.39/3.12/3.75

图注意力层数	MAE(15/30/60 min)	RMSE(15/30/60 min)	MAPE(15/30/60 min)	训练时间(epoch)/s
1	1.89/2.42/3.09	3.36/4.37/5.46	4.14/5.32/6.88	15
2	1.91/2.41/3.42	3.57/4.68/5.86	4.35/5.59/7.27	24
3	1.90/2.85/3.54	3.70/4.77/5.99	4.57/5.81/7.48	37

模型	参数量	训练时间(epoch)/s	测试时间/s
STGCN	582 783	55	51
STTN	483 421	269	257
AST-GCN	659 387	81	78

一种用于交通预测的注意力时空图神经网络

Attention spatial-temporal graph neural network for traffic prediction

RichHTML

PDF (PC)

赞

可视化

摘要/Abstract

引用本文

使用本文

图/表 10

参考文献 19

相关文章 15

Metrics

本文评价

推荐阅读 10

[1]	刘博翀, 蔡怀宇, 杨诗远, 李灏天, 汪毅, 陈晓冬. 一种用于自动驾驶场景的轻量级语义分割网络[J]. 西安电子科技大学学报, 2023, 50(1): 118-128.
[2]	张雅荔, 李文元, 李昌禄, 丁少博. 融合注意力引导的多尺度低照度图像增强方法[J]. 西安电子科技大学学报, 2023, 50(1): 129-136.
[3]	崔少国,陈思奇,杜兴. 面向目标情感分析的双重图注意力网络模型[J]. 西安电子科技大学学报, 2023, 50(1): 137-148.
[4]	张强, 杨欣朋, 赵世祥, 卫栋栋, 韩臻. 注意力机制的SAR图像车辆目标检测网络[J]. 西安电子科技大学学报, 2023, 50(1): 36-47.
[5]	张兆宇,田春娜,周恒,田西兰. 联合在线分类的双注意力RGBT孪生网络跟踪[J]. 西安电子科技大学学报, 2022, 49(6): 76-85.
[6]	刘侍刚,张同,杨建功,葛宝. 递进式空洞残差深度双目立体匹配网络[J]. 西安电子科技大学学报, 2022, 49(5): 175-180.
[7]	王侃, 王孟洋, 刘鑫, 田国强, 李川, 刘伟. 融合自注意力机制与CNN-BiGRU的事件检测[J]. 西安电子科技大学学报, 2022, 49(5): 181-188.
[8]	陈俊杰, 邓洪高, 马谋, 蒋俊正. GCN-GRU:一种无线传感器网络故障检测模型[J]. 西安电子科技大学学报, 2022, 49(5): 60-67.
[9]	高德勇,康自兵,王松,王阳萍. 利用卷积块注意力机制识别人体动作的方法[J]. 西安电子科技大学学报, 2022, 49(4): 144-155.
[10]	刘迪,郭继昌,汪昱东,张怡. 融合注意力机制的多尺度显著性目标检测网络[J]. 西安电子科技大学学报, 2022, 49(4): 118-126.
[11]	张静,张雪英,陈桂军,闫超. 结合3D-CNN和频-空注意力机制的EEG情感识别[J]. 西安电子科技大学学报, 2022, 49(3): 191-198.
[12]	马思珂,赵萌,石凡,孙续国,陈胜勇. 注意力驱动的细胞团簇细胞核分割算法[J]. 西安电子科技大学学报, 2022, 49(2): 198-206.
[13]	董如婵,焦李成,赵进,沈维燕. 一种深度融合机制的遥感图像目标检测技术[J]. 西安电子科技大学学报, 2021, 48(5): 128-138.
[14]	宋建锋,苗启广,王崇晓,徐浩,杨瑾. 注意力机制的多尺度单目标跟踪算法[J]. 西安电子科技大学学报, 2021, 48(5): 110-116.
[15]	成磊,王玥,田春娜. 一种添加残差注意力机制的视觉目标跟踪算法[J]. 西安电子科技大学学报, 2020, 47(6): 148-157.