Low-Rank Hankel Tensor Completion for Traffic Speed Estimation

被引:24
作者
Wang, Xudong [1 ]
Wu, Yuankai [1 ]
Zhuang, Dingyi [1 ]
Sun, Lijun [1 ]
机构
[1] McGill Univ, Dept Civil Engn, Montreal, PQ H3A 0C3, Canada
基金
加拿大自然科学与工程研究理事会; 加拿大创新基金会;
关键词
Spatio-temporal traffic data; traffic state esti-mation; missing data imputation; low-rank tensor completion; delay embedding transform; STATE ESTIMATION; MATRIX COMPLETION; FILTER; FLOW;
D O I
10.1109/TITS.2023.3247961
中图分类号
TU [建筑科学];
学科分类号
0813 ;
摘要
This paper studies the traffic state estimation (TSE) problem using sparse observations from mobile sensors. Most existing TSE methods either rely on well-defined physical traffic flow models or require large amounts of simulation data as input to train machine learning models. Different from previous studies, we propose a purely data-driven and model-free solution in this paper. We consider the TSE as a spatiotemporal matrix completion/interpolation problem, and apply spatiotemporal delay embedding to transform the original incomplete matrix into a fourth-order Hankel structured tensor. By imposing a low-rank assumption on this tensor structure, we can approximate and characterize both global and local spatiotemporal patterns in a data-driven manner. We use the truncated nuclear norm of a balanced spatiotemporal unfolding -- in which each column represents the vectorization of a small patch in the original matrix -- to approximate the tensor rank. An efficient solution algorithm based on the Alternating Direction Method of Multipliers (ADMM) is developed for model learning. The proposed framework only involves two hyperparameters, spatial and temporal window lengths, which are easy to set given the degree of data sparsity. We conduct numerical experiments on real-world high-resolution trajectory data, and our results demonstrate the effectiveness and superiority of the proposed model in some challenging scenarios.
引用
收藏
页码:4862 / 4871
页数:10
相关论文
共 34 条
[1]  
Boyd S., 2011, DISTRIBUTED OPTIMIZA
[2]   A nonconvex low-rank tensor completion model for spatiotemporal traffic data imputation [J].
Chen, Xinyu ;
Yang, Jinming ;
Sun, Lijun .
TRANSPORTATION RESEARCH PART C-EMERGING TECHNOLOGIES, 2020, 117
[3]  
Colyar James, 2007, Technical Report FHWA-HRT-07-030
[4]   Traffic state estimation based on Eulerian and Lagrangian observations in a mesoscopic modeling framework [J].
Duret, Aurelien ;
Yuan, Yufei .
TRANSPORTATION RESEARCH PART B-METHODOLOGICAL, 2017, 101 :51-71
[5]   Total-Variation-Regularized Low-Rank Matrix Factorization for Hyperspectral Image Restoration [J].
He, Wei ;
Zhang, Hongyan ;
Zhang, Liangpei ;
Shen, Huanfeng .
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, 2016, 54 (01) :176-188
[6]   Incorporation of Lagrangian measurements in freeway traffic state estimation [J].
Herrera, Juan C. ;
Bayen, Alexandre M. .
TRANSPORTATION RESEARCH PART B-METHODOLOGICAL, 2010, 44 (04) :460-481
[7]   Fast and Accurate Matrix Completion via Truncated Nuclear Norm Regularization [J].
Hu, Yao ;
Zhang, Debing ;
Ye, Jieping ;
Li, Xuelong ;
He, Xiaofei .
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 2013, 35 (09) :2117-2130
[8]   Physics Informed Deep Learning for Traffic State Estimation [J].
Huang, Jiheng ;
Agarwal, Shaurya .
2020 IEEE 23RD INTERNATIONAL CONFERENCE ON INTELLIGENT TRANSPORTATION SYSTEMS (ITSC), 2020,
[9]   Annihilating Filter-Based Low-Rank Hankel Matrix Approach for Image Inpainting [J].
Jin, Kyong Hwan ;
Ye, Jong Chul .
IEEE TRANSACTIONS ON IMAGE PROCESSING, 2015, 24 (11) :3498-3511
[10]   Tensor Decompositions and Applications [J].
Kolda, Tamara G. ;
Bader, Brett W. .
SIAM REVIEW, 2009, 51 (03) :455-500