Highway Travel Time Prediction Using Sparse Tensor Completion Tactics and K-Nearest Neighbor Pattern Matching Method

Remote transportation microwave sensor (RTMS) technology is being promoted for China’s highways. The distance is about 2 to 5 km between RTMSs, which leads to missing data and data sparseness problems. These two problems seriously restrict the accuracy of travel time prediction. Aiming at the data-m...

Full description

Saved in:
Bibliographic Details
Main Authors: Jiandong Zhao, Yuan Gao, Jinjin Tang, Lingxi Zhu, Jiaqi Ma
Format: Article
Language:English
Published: Wiley 2018-01-01
Series:Journal of Advanced Transportation
Online Access:http://dx.doi.org/10.1155/2018/5721058
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850176716878643200
author Jiandong Zhao
Yuan Gao
Jinjin Tang
Lingxi Zhu
Jiaqi Ma
author_facet Jiandong Zhao
Yuan Gao
Jinjin Tang
Lingxi Zhu
Jiaqi Ma
author_sort Jiandong Zhao
collection DOAJ
description Remote transportation microwave sensor (RTMS) technology is being promoted for China’s highways. The distance is about 2 to 5 km between RTMSs, which leads to missing data and data sparseness problems. These two problems seriously restrict the accuracy of travel time prediction. Aiming at the data-missing problem, based on traffic multimode characteristics, a tensor completion method is proposed to recover the lost RTMS speed and volume data. Aiming at the data sparseness problem, virtual sensor nodes are set up between real RTMS nodes, and the two-dimensional linear interpolation and piecewise method are applied to estimate the average travel time between two nodes. Next, compared with the traditional K-nearest neighbor method, an optimal KNN method is proposed for travel time prediction. optimization is made in three aspects. Firstly, the three original state vectors, that is, speed, volume, and time of the day, are subdivided into seven periods. Secondly, the traffic congestion level is added as a new state vector. Thirdly, the cross-validation method is used to calibrate the K value to improve the adaptability of the KNN algorithm. Based on the data collected from Jinggangao highway, all the algorithms are validated. The results show that the proposed method can improve data quality and prediction precision of travel time.
format Article
id doaj-art-bdb3e590bcb6410d904c7aae64d4cb7d
institution OA Journals
issn 0197-6729
2042-3195
language English
publishDate 2018-01-01
publisher Wiley
record_format Article
series Journal of Advanced Transportation
spelling doaj-art-bdb3e590bcb6410d904c7aae64d4cb7d2025-08-20T02:19:12ZengWileyJournal of Advanced Transportation0197-67292042-31952018-01-01201810.1155/2018/57210585721058Highway Travel Time Prediction Using Sparse Tensor Completion Tactics and K-Nearest Neighbor Pattern Matching MethodJiandong Zhao0Yuan Gao1Jinjin Tang2Lingxi Zhu3Jiaqi Ma4School of Traffic and Transportation, Beijing Jiaotong University, Beijing, ChinaSchool of Mechanical and Electronic Control Engineering, Beijing Jiaotong University, Beijing, ChinaSchool of Traffic and Transportation, Beijing Jiaotong University, Beijing, ChinaSchool of Traffic and Transportation, Beijing Jiaotong University, Beijing, ChinaDepartment of Civil and Architectural Engineering and Construction Management, University of Cincinnati, Cincinnati, OH, USARemote transportation microwave sensor (RTMS) technology is being promoted for China’s highways. The distance is about 2 to 5 km between RTMSs, which leads to missing data and data sparseness problems. These two problems seriously restrict the accuracy of travel time prediction. Aiming at the data-missing problem, based on traffic multimode characteristics, a tensor completion method is proposed to recover the lost RTMS speed and volume data. Aiming at the data sparseness problem, virtual sensor nodes are set up between real RTMS nodes, and the two-dimensional linear interpolation and piecewise method are applied to estimate the average travel time between two nodes. Next, compared with the traditional K-nearest neighbor method, an optimal KNN method is proposed for travel time prediction. optimization is made in three aspects. Firstly, the three original state vectors, that is, speed, volume, and time of the day, are subdivided into seven periods. Secondly, the traffic congestion level is added as a new state vector. Thirdly, the cross-validation method is used to calibrate the K value to improve the adaptability of the KNN algorithm. Based on the data collected from Jinggangao highway, all the algorithms are validated. The results show that the proposed method can improve data quality and prediction precision of travel time.http://dx.doi.org/10.1155/2018/5721058
spellingShingle Jiandong Zhao
Yuan Gao
Jinjin Tang
Lingxi Zhu
Jiaqi Ma
Highway Travel Time Prediction Using Sparse Tensor Completion Tactics and K-Nearest Neighbor Pattern Matching Method
Journal of Advanced Transportation
title Highway Travel Time Prediction Using Sparse Tensor Completion Tactics and K-Nearest Neighbor Pattern Matching Method
title_full Highway Travel Time Prediction Using Sparse Tensor Completion Tactics and K-Nearest Neighbor Pattern Matching Method
title_fullStr Highway Travel Time Prediction Using Sparse Tensor Completion Tactics and K-Nearest Neighbor Pattern Matching Method
title_full_unstemmed Highway Travel Time Prediction Using Sparse Tensor Completion Tactics and K-Nearest Neighbor Pattern Matching Method
title_short Highway Travel Time Prediction Using Sparse Tensor Completion Tactics and K-Nearest Neighbor Pattern Matching Method
title_sort highway travel time prediction using sparse tensor completion tactics and k nearest neighbor pattern matching method
url http://dx.doi.org/10.1155/2018/5721058
work_keys_str_mv AT jiandongzhao highwaytraveltimepredictionusingsparsetensorcompletiontacticsandknearestneighborpatternmatchingmethod
AT yuangao highwaytraveltimepredictionusingsparsetensorcompletiontacticsandknearestneighborpatternmatchingmethod
AT jinjintang highwaytraveltimepredictionusingsparsetensorcompletiontacticsandknearestneighborpatternmatchingmethod
AT lingxizhu highwaytraveltimepredictionusingsparsetensorcompletiontacticsandknearestneighborpatternmatchingmethod
AT jiaqima highwaytraveltimepredictionusingsparsetensorcompletiontacticsandknearestneighborpatternmatchingmethod