关键节点感知与时空自适应的Transformer交通流预测方法

doi:10.3778/j.issn.1002-8331.2502-0194

摘要/Abstract

摘要： 交通流数据本身具有明显的时空特性，不仅与时间和位置密切相关，还与道路网络的空间结构息息相关。目前的研究方法通常依赖于复杂的神经网络来捕捉交通流的时空相关性，但往往未能深入挖掘交通流数据的内在属性。此外，现有方法对于嵌入层的数据处理和交通流的时空特征也缺乏充分的探讨。为了解决这些问题，提出了一种名为KASTFormer的模型。该模型通过关键节点感知空间自注意力机制，有效地捕捉交通流枢纽道路的特征，并结合时间、语义和地理自注意力机制，进一步加强了短期和长期时空依赖的建模。同时，设计了一种时空自适应嵌入方法，通过统一建模时空关系捕捉交通流数据及其时空特征。通过在三个数据集上的实验验证了模型的有效性，并通过消融实验和模型结构性能分析进一步证明了其实际效果。

关键词: 关键节点感知, 时空自适应, 注意力机制, 交通预测, 时空特征

Abstract: Traffic flow data itself has obvious spatial-temporal characteristics, which are closely related to not only time and location, but also the spatial structure of the road network. Current research methods usually rely on complex neural networks to capture the spatial-temporal correlation of traffic flow, but often fail to deeply mine the intrinsic properties of traffic flow data. In addition, the existing methods lack sufficient discussion on the data processing of the embedding layer and the spatial-temporal characteristics of traffic flow. To address these issues, a model called KASTFormer is proposed. The model effectively captures the characteristics of traffic flow hub roads through the spatial self-attention mechanism sensed by key nodes, and further strengthens the modeling of short-term and long-term spatial-temporal dependencies by combining temporal, semantic and geographic self-attention mechanisms. At the same time, a spatial-temporal adaptive embedding method is designed to capture the traffic flow data and its spatial-temporal characteristics through a unified modeling spatial-temporal relationship. Finally, the effectiveness of the model is verified by experiments on three data sets, and the practical effect is further proved by ablation experiments and model structure performance analysis.

Key words: key-node awareness, spatial-temporal adaptation, attention mechanism, traffic prediction, spatial-temporal features

李卫军, 熊章友, 杨国梁, 朱晓娟, 马馨瑜. 关键节点感知与时空自适应的Transformer交通流预测方法[J]. 计算机工程与应用, 2026, 62(8): 380-391.

LI Weijun, XIONG Zhangyou, YANG Guoliang, ZHU Xiaojuan, MA Xinyu. Transformer Traffic Flow Prediction Method Based on Key Node Perception and Spatial-Temporal Adaptation[J]. Computer Engineering and Applications, 2026, 62(8): 380-391.

参考文献

[1] ZANG D，LING J W，WEI Z H，et al.Long-term traffic speed prediction based on multiscale spatio?temporal feature learning network[J].IEEE Transactions on Intelligent Transportation Systems，2019，20（10）：3700-3709.
[2] BOGAERTS T，MASEGOSA A D，ANGARITA-ZAPATA J S，et al.A graph CNN-LSTM neural network for short and long-term traffic forecasting based on trajectory data[J].Transportation Research Part C：Emerging Technologies，2020，112：62-77.
[3] LI Y G，YU R，SHAHABI C，et al.Diffusion convolutional recurrent neural network：data-driven traffic forecasting[EB/OL].[2025-01-09].https：//arxiv.org/abs/1707.01926.
[4] ZHENG C P，FAN X L，WANG C，et al.GMAN：a graph multi-attention network for traffic prediction[C]//Proceedings of the 34th AAAI Conference on Artificial Intelligence.Palo Alto：AAAI，2020：1234-1241.
[5] GUO S N，LIN Y F，FENG N，et al.Attention based spatial-temporal graph convolutional networks for traffic flow forecasting[C]//Proceedings of the 33rd AAAI Conference on Artificial Intelligence and 31st Innovative Applications of Artificial Intelligence Conference and 9th AAAI Symposium on Educational Advances in Artificial Intelligence.New York：ACM，2019：922-929.
[6] WEI Z B，ZHAO H X，LI Z S，et al.STGSA：a novel spatial-temporal graph synchronous aggregation model for traffic prediction[J].IEEE/CAA Journal of Automatica Sinica，2023，10（1）：226-238.
[7] WANG Y，ZHENG J，DU Y Q，et al.Traffic-GGNN：predicting traffic flow via attentional spatial-temporal gated graph neural networks[J].IEEE Transactions on Intelligent Transportation Systems，2022，23（10）：18423-18432.
[8] LI M Z，ZHU Z X.Spatial-temporal fusion graph neural networks for traffic flow forecasting[C]//Proceedings of the 35th AAAI Conference on Artificial Intelligence.Palo Alto：AAAI，2021：4189-4196.
[9] BAI J D，ZHU J W，SONG Y J，et al.A3T-GCN：attention temporal graph convolutional network for traffic forecasting[J].ISPRS International Journal of Geo-Information，2021，10（7）：485.
[10] ALI A，ZHU Y M，ZAKARYA M.Exploiting dynamic spatio-temporal graph convolutional neural networks for citywide traffic flows prediction[J].Neural Networks，2022，145：233-247.
[11] CHEN C L，LIU Y B，CHEN L，et al.Bidirectional spatial-temporal adaptive Transformer for urban traffic flow forecasting[J].IEEE Transactions on Neural Networks and Learning Systems，2023，34（10）：6913-6925.
[12] LIU J S，KANG Y，LI H，et al.STGHTN：spatial-temporal gated hybrid Transformer network for traffic flow forecasting[J].Applied Intelligence，2023，53（10）：12472-12488.
[13] SHAO Z Z，ZHANG Z，WANG F，et al.Spatial-temporal identity：a simple yet effective baseline for multivariate time series forecasting[C]//Proceedings of the 31st ACM International Conference on Information & Knowledge Management.New York：ACM，2022：4454-4458.
[14] 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.
[15] CHOROWSKI J，BAHDANAU D，SERDYUK D，et al.Attention-based models for speech recognition[C]//Advances in Neural Information Processing Systems 28，2015：577-585.
[16] WEI G.A summary of traffic flow forecasting methods[J].Journal of Highway and Transportation Research and Development，2004，21（3）：82-85.
[17] ZHENG H F，LIN F，FENG X X，et al.A hybrid deep learning model with attention-based conv-LSTM networks for short-term traffic flow prediction[J].IEEE Transactions on Intelligent Transportation Systems，2021，22（11）：6910-6920.
[18] DRUCKER H，BURGES C J C，KAUFMAN L，et al.Support vector regression machines[C]//Advances in Neural Information Processing Systems 9，1996：155-161.
[19] BREIMAN L.Random forests[J].Machine Learning，2001，45（1）：5-32.
[20] ZHANG J B，ZHENG Y，QI D K.Deep spatio-temporal residual networks for citywide crowd flows prediction[C]//Proceedings of the 31st AAAI Conference on Artificial Intelligence.Palo Alto：AAAI，2017，31（1）：1655-1661.
[21] SUN P，BOUKERCHE A，TAO Y J.SSGRU：a novel hybrid stacked GRU-based traffic volume prediction approach in a road network[J].Computer Communications，2020，160：502-511.
[22] YANG H，JIANG C，SONG Y，et al.TARGCN：temporal attention recurrent graph convolutional neural network for traffic prediction[J].Complex & Intelligent Systems，2024，10（6）：8179-8196.
[23] JI J H，WANG J Y，HUANG C，et al.Spatio-temporal self-supervised learning for traffic flow prediction[C]//Proceedings of the 37th AAAI Conference on Artificial Intelligence.Palo Alto：AAAI，2023：4356-4364.
[24] QI X D，YAO J F，WANG P，et al.Combining weather factors to predict traffic flow：a spatial-temporal fusion graph convolutional network-based deep learning approach[J].IET Intelligent Transport Systems，2024，18（3）：528-539.
[25] CHEN L，SHAO W，LV M Q，et al.AARGNN：an attentive attributed recurrent graph neural network for traffic flow prediction considering multiple dynamic factors[J].IEEE Transactions on Intelligent Transportation Systems，2022，23（10）：17201-17211.
[26] 王庆荣，田可可，朱昌锋，等.融合多因素的短时交通流预测研究[J].计算机工程与应用，2022，58（21）：309-316.
WANG Q R，TIAN K K，ZHU C F，et al.Short term traffic flow prediction based on multi-factors[J].Computer Engineering and Applications，2022，58（21）：309-316.
[27] JIANG R H，WANG Z N，YONG J W，et al.Spatio-temporal meta-graph learning for traffic forecasting[C]//Proceedings of the 37th AAAI Conference on Artificial Intelligence.Palo Alto：AAAI，2023：8078-8086.
[28] 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 34th AAAI Conference on Artificial Intelligence.Palo Alto：AAAI，2020：914-921.
[29] ZHANG L，WU J Q，SHEN J，et al.SATP-GAN：self-attention based generative adversarial network for traffic flow prediction[J].Transportmetrica B：Transport Dynamics，2021，9（1）：552-568.
[30] ZHANG Q Y，CHANG W F，YIN C H，et al.Attention-based spatial-temporal convolution gated recurrent unit for traffic flow forecasting[J].Entropy，2023，25（6）：938.
[31] HUANG X，ZHANG B W，FENG S S，et al.Interpretable local flow attention for multi-step traffic flow prediction[J].Neural Networks，2023，161：25-38.
[32] YU Q，DING W L，ZHANG H，et al.Rethinking attention mechanism for spatio-temporal modeling：a decoupling perspective in traffic flow prediction[C]//Proceedings of the 33rd ACM International Conference on Information and Knowledge Management.New York：ACM，2024：3032-3041.
[33] HUO G Y，ZHANG Y，WANG B Y，et al.Hierarchical spatio-temporal graph convolutional networks and transformer network for traffic flow forecasting[J].IEEE Transactions on Intelligent Transportation Systems，2023，24（4）：3855-3867.
[34] LI G Y，ZHAO Z L，GUO X G，et al.Towards integrated and fine?grained traffic forecasting：a spatio?temporal heterogeneous graph Transformer approach[J].Information Fusion，2024，102：102063.
[35] KONG J L，FAN X M，ZUO M，et al.ADCT-Net：adaptive traffic forecasting neural network via dual-graphic cross-fused transformer[J].Information Fusion，2024，103：102122.
[36] JIANG J W，HAN C K，ZHAO W X，et al.PDFormer：propagation delay-aware dynamic long-range transformer for traffic flow prediction[C]//Proceedings of the 37th AAAI Conference on Artificial Intelligence.Palo Alto：AAAI，2023：4365-4373.
[37] YU B，YIN H T，ZHU Z X.Spatio-temporal graph convolutional networks：a deep learning framework for traffic forecasting[C]//Proceedings of the 27th International Joint Conference on Artificial Intelligence.New York：ACM，2018：3634-3640.
[38] WU Z H，PAN S R，LONG G D，et al.Graph WaveNet for deep spatial-temporal graph modeling[C]//Proceedings of the 28th International Joint Conference on Artificial Intelligence，2019：1907-1913.
[39] FANG Z，LONG Q Q，SONG G J，et al.Spatial-temporal graph ODE networks for traffic flow forecasting[C]//Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining.New York：ACM，2021：364-373.
[40] LAN S Y，MA Y T，HUANG W，et al.DSTAGNN：dynamic spatial-temporal aware graph neural network for traffic flow forecasting[C]//Proceedings of the 39th International Conference on Machine Learning，2022：11906-11917.