基于时空图注意力网络的交通流量预测
Spatial-Temporal Graph Attention Networks for Traffic Flow Forecasting

DOI: 10.12677/CSA.2021.1111281, PP. 2770-2779

Keywords: 智能交通，交通流量预测，时空相关性，图注意力网络
Intelligent Transportation, Traffic Flow Forecasting, Spatial-Temporal Correlations, Graph Attention Networks

Full-Text   Cite this paper   Add to My Lib

Abstract:

准确的交通流预测对于提高智能交通系统的安全性、稳定性和效率至关重要。然而，考虑到交通流之间复杂的空间和时间依赖性，对交通网络中的时空关系进行建模是一项具有挑战性的任务。本文设计了一种新颖的时空图注意力网络(STGAL)来提取交通流动态和静态的时空相关性，更加有效解决交通流量预测问题。具体地说，它主要有三个模块：1) 图注意力网络用于捕捉节点之间的空间关系，对邻域节点进行有区别的信息聚合；2) 长短期记忆网络用于捕获交通流的时间相关性；3) 空间和时间注意力网络用于捕捉交通流中动态变化的时空关系。我们不仅考虑邻域节点的特征和边的权重来生成新的节点表示，而且考虑交通流动态变化的时空模式。此外，为了挖掘周期性数据对预测任务的影响，我们融合了交通流最近、日周期和周周期的三个组件特征信息。在多步交通预测任务上的大量实验证明了STGAL的有效性和优越性。
Accurate traffic flow forecasting is critical in improving safety, stability, and efficiency of intelligent transportation systems. However, considering the complex spatial and temporal dependence between traffic flows, modeling the spatial-temporal correlation in traffic is a challenging task. In this paper, we design a novel spatial-temporal graph attention networks (STGAL) to extract the dynamic and static spatial-temporal correlation of traffic flow simultaneously, and effectively address the problem of traffic flow forecasting. Specifically, there are three main modules: 1) Graph attention network is used to capture the spatial correlation between nodes and to aggregate the information of the neighborhood nodes differently; 2) Long short-term memory network to capture the tem-poral correlation of traffic flow; 3) Spatial and temporal attention networks to capture the spatial-temporal correlation of dynamic changes in traffic flow. We consider the characteristics of neighborhood nodes, weights of edges and spatial-temporal pattern of traffic flow dynamics. In addition, we integrate the recent, daily, and weekly component feature information of traffic flow to mine the impact of periodic data on prediction tasks. A large number of experiments on multi-step traffic forecasting tasks have proved the effectiveness and superiority of STGAL.

References

[1]	赵娜, 袁家斌, 徐晗. 智能交通系统综述[J]. 计算机科学, 2014, 41(11): 7-11+45.
[2]	杨春霞, 秦家鹏, 王庆, 李欣栩. 基于多车道加权融合的短时交通流预测研究[J]. 公路交通科技, 2021, 38(1): 121-127.
[3]	Lippi, M., Ber-tini, M. and Frasconi, P. (2013) Short-Term Traffic Flow Forecasting: An Experimental Comparison of Time-Series Analysis and Supervised Learning. IEEE Transactions on Intelligent Transportation Systems, 14, 871-882. https://doi.org/10.1109/TITS.2013.2247040
[4]	李磊, 张青苗, 赵军辉, 聂逸文. 基于改进CNN-LSTM组合模型的分时段短时交通流预测[J]. 应用科学学报, 2021, 39(2): 185-198.
[5]	Li, Y., Yu, R., Shahabi, C. and Liu, Y. (2018) Diffusion Convolutional Recurrent Neural Network: Data-Driven Traffic Forecasting. International Conference on Learning Representations, Vancouver, 30 April-3 May 2018, 1-10. https://openreview.net/forum?id=SJiHXGWAZ
[6]	Yu, B., Yin, H. and Zhu, Z. (2018) Spatial-Temporal Graph Convolutional Networks: A Deep Learning Framework for Traffic Forecasting. Proceedings of the 27th International Joint Conference on Artificial Intelligence, Stockholm, 13-19 July 2018, 3634-3640. https://doi.org/10.24963/ijcai.2018/505
[7]	Wu, Z., Pan, S., Long, G., Jiang, J. and Zhang, C. (2019) Graph Wave Net for Deep Spatial-Temporal Graph Modeling. Proceedings of the 28th International Joint Conference on Artificial In-telligence, Macao, 10-16 August 2019, 1907-1913. https://doi.org/10.24963/ijcai.2019/264
[8]	Guo, S., Lin, Y., Feng, N., Song, C. and Wan, H. (2019) Attention Based Spatial-Temporal Graph Convolutional Networks for Traffic Flow Forecasting. Proceedings of the AAAI Conference on Artificial Intelligence, 33, 922-929. https://doi.org/10.1609/aaai.v33i01.3301922
[9]	Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A.N., Kaiser, L. and Polosukhin, I. (2017) Attention Is All You Need. Proceedings of the 31st International Conference on Neural Information Processing Systems, Long Beach, 4-9 December 2017, 5998-6008.
[10]	Veli？kovi？, P., Cucurull, G., Casanova, A., Romero, A., Lio, P. and Bengio, Y. (2018) Graph Attention Networks. International Conference on Learning Representations, Vancouver, 30 April-3 May 2018, 1-12. https://openreview.net/forum?id=rJXMpikCZ
[11]	Feng, X., Guo, J., Qin, B., Liu, T. and Liu, Y. (2017) Effective Deep Memory Networks for Distant Supervised Relation Extraction. Proceedings of the 26th International Joint Con-ference on Artificial Intelligence, Melbourne, 19-25 August 2017, 4002-4008. https://doi.org/10.24963/ijcai.2017/559

Full-Text