基于时空异质图卷积的交通流量预测
Spatio-Temporal Heterogeneous Graph-Based Convolutional Networks for Traffic Flow Forecasting
DOI: 10.12677/CSA.2023.133038, PDF,    国家自然科学基金支持
作者: 辛笑阳, 程泽生*：青岛大学计算机科学技术学院，山东 青岛；吕 阳, 王晓彤, 祁洋洋：青岛大学泛在网络与城市计算研究所，山东 青岛
关键词: 交通流量预测；异质图；深度学习；图卷积；Traffic Flow Forecasting； Heterogeneous Graphs； Deep Learning； Graph Convolution
摘要: 交通流量预测在智慧交通建设中发挥着至关重要的作用。为了充分挖掘交通网络中结点之间的空间相关性，本文提出一种基于异质图的深度时空模型STREGCN。首先，本文提出将交通网络抽象为异质图，增强图的表达能力，进而充分捕获交通网络的空间相关性。其次，本文采用基于线性门控单元的一维因果卷积去充分提取交通流量的时间相关性。最后，本文设计时空卷积模块的输出经过全连接层获得最终的交通流量预测结果。本文在开源的交通数据集PEMSD8进行了预测区间为5分钟和30分钟的交通流量预测实验。实验结果表明STREGCN模型与大多数基线模型相比，在未来短期和长期的交通流量预测任务上都有更好的表现。
Abstract: Traffic flow forecasting plays a crucial role in the construction of intelligent transportation. In order to fully exploit the spatial correlation between nodes in a traffic network, this paper proposes a deep spatio-temporal model STREGCN based on heterogeneous graphs. Firstly, this paper proposes to abstract the traffic network as a heterogeneous graph to enhance the graph’s expressiveness, and then fully capture the spatial correlation of the traffic network. Secondly, this paper uses one-dimensional causal convolution based on linear gating units to fully extract the temporal correlation of traffic flows. Finally, this paper designs the output of the spatio-temporal convolution module to obtain the final traffic flow prediction results after a fully connected layer. In this paper, traffic flow prediction experiments with prediction intervals of 5 and 30 minutes are conducted on the open-source traffic dataset PEMSD8. The experimental results show that the STREGCN model performs better than most baseline models for both short- and long-term future traffic flow forecasting tasks.
文章引用：辛笑阳, 程泽生, 吕阳, 王晓彤, 祁洋洋. 基于时空异质图卷积的交通流量预测[J]. 计算机科学与应用, 2023, 13(3): 399-409. https://doi.org/10.12677/CSA.2023.133038

参考文献

[1] Cheng, Z., Jian, S., Rashidi, T.H., et al. (2020) Integrating Household Travel Survey and Social Media Data to Improve the Quality of Od Matrix: A Comparative Case Study. IEEE Transactions on Intelligent Transportation Systems, 21, 2628-2636. [Google Scholar] [CrossRef] 
[2] Tang, J., Chen, X., Hu, Z., et al. (2019) Traffic Flow Prediction Based on Combination of Support Vector Machine and Data Denoising Schemes. Physica A: Statistical Me-chanics and Its Applications, 534, Article ID: 120642. [Google Scholar] [CrossRef] 
[3] Feng, X., Ling, X., Zheng, H., et al. (2018) Adaptive Mul-ti-Kernel SVM with Spatial-Temporal Correlation for Short-Term Traffic Flow Prediction. IEEE Transactions on Intelli-gent Transportation Systems, 20, 2001-2013. [Google Scholar] [CrossRef] 
[4] Lv, Z., Li, J., Dong, C., et al. (2021) Deep Learning in the COVID-19 Epidemic: A Deep Model for Urban Traffic Revitalization Index. Data & Knowledge Engineering, 135, Arti-cle ID: 101912. [Google Scholar] [CrossRef] [PubMed]
[5] Li, H., Lv, Z., Li, J., et al. (2022) Traffic Flow Forecasting in the COVID-19: A Deep Spatial-Temporal Model Based on Discrete Wavelet Transformation. ACM Transactions on Knowledge Discovery from Data, 17, 1-28. [Google Scholar] [CrossRef] 
[6] Lv, Z., Li, J., Dong, C., et al. (2021) DeepSTF: A Deep Spatial-Temporal Forecast Model of Taxi Flow. The Computer Journal. [Google Scholar] [CrossRef] 
[7] Wang, Y., Lv, Z., Sheng, Z., et al. (2022) A Deep Spatio-Temporal Meta-Learning Model for Urban Traffic Revitalization Index Predic-tion in the COVID-19 Pandemic. Advanced Engineering Informatics, 53, Article ID: 101678. [Google Scholar] [CrossRef] [PubMed]
[8] Wang, Y., Zhao, A., Li, J., et al. (2022) Multi-Attribute Graph Convolution Network for Regional Traffic Flow Prediction. Neural Processing Letters, 1-27. [Google Scholar] [CrossRef] 
[9] Cheng, Z., Rashidi, T.H., Jian, S., et al. (2022) A Spa-tio-Temporal Autocorrelation Model for Designing a Carshare System Using Historical Heterogeneous Data: Policy Suggestion. Transportation Research Part C: Emerging Technologies, 141, Article ID: 103758. [Google Scholar] [CrossRef] 
[10] Xu, Z., Lv, Z., Li, J., et al. (2022) A Novel Perspective on Travel Demand Prediction Considering Natural Environmental and Socioeconomic Factors. IEEE Intelligent Transportation Systems Magazine, 15, 136-159. [Google Scholar] [CrossRef] 
[11] Sun, H., Lv, Z., Li, J., et al. (2022) Prediction of Cancellation Probability of Online Car-Hailing Orders Based on Multi-Source Heterogeneous Data Fusion. Wireless Algorithms, Sys-tems, and Applications: 17th International Conference, WASA 2022, Dalian, 24-26 November 2022, 168-180. [Google Scholar] [CrossRef] 
[12] Song, D.M. and Zhang, Y.H. (2022) Opinion Formation on a Time Varying Dynamic Network with Different Personality Types: Stubborn, Follower, and Extreme. 2022 5th Interna-tional Conference on Big Data Technologies (ICBDT 2022), Qingdao, 23-25 September 2022, 190-195. [Google Scholar] [CrossRef] 
[13] Yuan, G., Li, J., Lv, Z., et al. (2021) DDCAttNet: Road Segmenta-tion Network for Remote Sensing Images. Wireless Algorithms, Systems, and Applications: 16th International Confer-ence, WASA 2021, Nanjing, 25-27 June 2021, 457-468. [Google Scholar] [CrossRef] 
[14] Sun, H., Lv, Z., Li, J., et al. (2023) Will the Order Be Canceled? Order Cancellation Probability Prediction Based on Deep Re-sidual Model. Transportation Research Record. [Google Scholar] [CrossRef] 
[15] Jiang, W. and Zhang, L. (2018) Geospatial Data to Images: A Deep-Learning Framework for Traffic Forecasting. Tsinghua Science and Technology, 24, 52-64. [Google Scholar] [CrossRef] 
[16] Chen, F., Chen, Z., Biswas, S., et al. (2020) Graph Convolutional Networks with Kalman Filtering for Traffic Prediction. Proceedings of the 28th Internation-al Conference on Advances in Geographic Information Systems, Washington DC, 3-6 November 2020, 135-138. [Google Scholar] [CrossRef] 
[17] Zhao, L., Song, Y., Zhang, C., et al. (2019) T-gcn: A Temporal Graph Convolutional Network for Traffic Prediction. IEEE Transactions on Intelligent Transportation Systems, 21, 3848-3858. [Google Scholar] [CrossRef] 
[18] Yu, B., Yin, H. and Zhu, Z. (2017) Spatio-Temporal Graph Convolutional Networks: A Deep Learning Framework for Traffic Forecasting. [Google Scholar] [CrossRef] 
[19] Kumar, S.V. and Vanajakshi, L. (2015) Short-Term Traffic Flow Pre-diction Using Seasonal ARIMA Model with Limited Input Data. European Transport Research Review, 7, 1-9. [Google Scholar] [CrossRef] 
[20] Kumar, S.V. (2017) Traffic Flow Prediction Using Kalman Fil-tering Technique. Procedia Engineering, 187, 582-587. [Google Scholar] [CrossRef] 
[21] Zhang, L., Liu, Q., Yang, W., et al. (2013) An Improved k-Nearest Neighbor Model for Short-Term Traffic Flow Prediction. Procedia—Social and Behavioral Sciences, 96, 653-662. [Google Scholar] [CrossRef] 
[22] Xu, Z., Li, J., Lv, Z., et al. (2021) A Graph Spa-tial-Temporal Model for Predicting Population Density of Key Areas. Computers & Electrical Engineering, 93, Article ID: 107235. [Google Scholar] [CrossRef] [PubMed]
[23] Lv, Z., Li, J., Xu, Z., et al. (2021) Parallel Computing of Spatio-Temporal Model Based on Deep Reinforcement Learning. Wireless Algorithms, Systems, and Ap-plications: 16th International Conference, WASA 2021, Nanjing, 25-27 June 2021, 391-403. [Google Scholar] [CrossRef] 
[24] Lv, Z., Li, J., Dong, C., et al. (2020) A Deep Spatial-Temporal Network for Vehicle Trajectory Prediction. Wireless Algorithms, Systems, and Applications: 15th International Confer-ence, WASA 2020, Qingdao, 13-15 September 2020, 359-369. [Google Scholar] [CrossRef] 
[25] Xu, Z., Lv, Z., Li, J., et al. (2022) A Novel Approach for Pre-dicting Water Demand with Complex Patterns Based on Ensemble Learning. Water Resources Management, 36, 4293-4312. [Google Scholar] [CrossRef] 
[26] Lv, Z., Li, J., Li, H., et al. (2021) Blind Travel Pre-diction Based on Obstacle Avoidance in Indoor Scene. Wireless Communications and Mobile Computing, 2021, Article ID: 5536386. [Google Scholar] [CrossRef] 
[27] Liang, Y., Li, Y., Guo, J., et al. (2022) Resource Compe-tition in Blockchain Networks under Cloud and Device Enabled Participation. IEEE Access, 10, 11979-11993. [Google Scholar] [CrossRef] 
[28] Lv, Z., Li, J., Dong, C., et al. (2021) DeepPTP: A Deep Pe-destrian Trajectory Prediction Model for Traffic Intersection. KSII Transactions on Internet & Information Systems, 15, 2321-2338. [Google Scholar] [CrossRef] 
[29] Sheng, Z., Lv, Z., Li, J., et al. (2023) Taxi Travel Time Prediction Based on Fusion of Traffic Condition Features. Computers and Electrical Engineering, 105, Article ID: 108530. [Google Scholar] [CrossRef] 
[30] Ye, R., Xu, Z. and Pang, J. (2022) DDFM: A Novel Perspective on Urban Travel Demand Forecasting Based on the Ensemble Empirical Mode Decomposition and Deep Learning. Proceedings of the 5th International Conference on Big Data Technologies, Qingdao, 23-25 September 2022, 373-379. [Google Scholar] [CrossRef] 
[31] Kipf, T.N. and Welling, M. (2016) Semi-Supervised Classification with Graph Convolutional Networks.
[32] Bruna, J., Zaremba, W., Szlam, A., et al. (2013) Spectral Networks and Locally Connected Networks on Graphs.
[33] Defferrard, M., Bresson, X. and Vander-gheynst, P. (2016) Convolutional Neural Networks on Graphs with Fast Localized Spectral Filtering. Advances in Neural Information Processing Systems 29: Annual Conference on Neural Information Processing Systems 2016, Barcelona, 5-10 December 2016, 3844-3852.
[34] Sun, H., Yang, C., Deng, L., et al. (2021) Periodicmove: Shift-Aware Human Mobility Recovery with Graph Neural Network. Proceedings of the 30th ACM International Conference on Information & Knowledge Management, 1-5 November 2021, 1734-1743. [Google Scholar] [CrossRef] 
[35] Cai, L., Janowicz, K., Mai, G., et al. (2020) Traffic Transformer: Capturing the Continuity and Periodicity of Time Series for Traffic Forecasting. Transactions in GIS, 24, 736-755. [Google Scholar] [CrossRef] 
[36] Yang, S., Ma, W., Pi, X. and Qian, S. (2019) A Deep Learning Approach to Real-Time Parking Occupancy Prediction in Transportation Networks Incorporating Multiple Spatio-Temporal Data Sources. Transportation Research Part C: Emerging Technolo-gies, 107, 248-265. [Google Scholar] [CrossRef] 
[37] Hong, H., Lin, Y., Yang, X., Li, Z., Fu, K., Wang, Z. and Ye, J. (2020) Heteta: Heterogeneous Information Network Embedding for Estimating Time of Arrival. Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, 6-10 July 2020, 2444-2454. [Google Scholar] [CrossRef] 
[38] Wang, M.Y. (2019) Deep Graph Library: To-wards Efficient and Scalable Deep Learning on Graphs. ICLR Workshop on Representation Learning on Graphs and Manifolds, New Orleans, 6 - 9 May 2019.