TY - JOUR
T1 - TST-Trans
T2 - A Transformer Network for Urban Traffic Flow Prediction
AU - Zhang, Ke
AU - Ren, Hongjin
AU - Kang, Jinbiao
AU - Guo, Cai
AU - Chen, Weiming
AU - Tao, Ming
AU - Dai, Hong Ning
AU - Wan, Shaohua
AU - Bao, Haiyong
N1 - Funding Information:
This work was supported in part by the National Natural Science Foundation of China under Grant 62173066 and 62072404, in part by the Science and Technology Plan Program of Huzhou City under Grant 2022GZ03, and in part by the Shanghai Natural Science Foundation under Grant 23ZR1417700.
Publisher Copyright:
© 2024 IEEE.
PY - 2024/11/18
Y1 - 2024/11/18
N2 - A critical challenge for predicting urban traffic flows is to simultaneously process time series and spatial features from heterogeneous traffic data collected by diverse Internet of Things (IoT) devices. Despite the advent of Transformer-based models with an advanced network structure and excellent prediction performance, standard Transformer models are still struggling to combine both spatial information and temporal relations of traffic flows. To address these challenges, we design a novel Transformer network, namely TST-Trans, for traffic flow prediction with high accuracy. In particular, we use learnable position encoders to replace traditional fixed position encoders. Meanwhile, we introduce a spatio-temporal embedding method that integrates temporal relationships and spatial information with external inputs, thereby capturing the spatio-temporal dependencies of traffic flows. Experiments with real-world datasets demonstrate that our proposed TST-Trans achieves better prediction accuracy than state-of-the-art methods while requiring fewer parameters.The research results increased by more than 10% compared with tranformer. Compared to STD-NET, there is a 2% to 10% improvement in performance on different datasets.
AB - A critical challenge for predicting urban traffic flows is to simultaneously process time series and spatial features from heterogeneous traffic data collected by diverse Internet of Things (IoT) devices. Despite the advent of Transformer-based models with an advanced network structure and excellent prediction performance, standard Transformer models are still struggling to combine both spatial information and temporal relations of traffic flows. To address these challenges, we design a novel Transformer network, namely TST-Trans, for traffic flow prediction with high accuracy. In particular, we use learnable position encoders to replace traditional fixed position encoders. Meanwhile, we introduce a spatio-temporal embedding method that integrates temporal relationships and spatial information with external inputs, thereby capturing the spatio-temporal dependencies of traffic flows. Experiments with real-world datasets demonstrate that our proposed TST-Trans achieves better prediction accuracy than state-of-the-art methods while requiring fewer parameters.The research results increased by more than 10% compared with tranformer. Compared to STD-NET, there is a 2% to 10% improvement in performance on different datasets.
KW - Internet of Things
KW - Spatio-temporal data
KW - Traffic prediction
KW - Transformer
UR - http://www.scopus.com/inward/record.url?scp=85210020796&partnerID=8YFLogxK
U2 - 10.1109/JIOT.2024.3501294
DO - 10.1109/JIOT.2024.3501294
M3 - Journal article
AN - SCOPUS:85210020796
SN - 2327-4662
JO - IEEE Internet of Things Journal
JF - IEEE Internet of Things Journal
M1 - 10756574
ER -