東華大學圖書館 |

語系: 繁體中文

說明(常見問題)

回圖書館首頁

手機版館藏查詢

登入

回首頁

切換: 標籤 | MARC模式 | ISBD

Traffic State Estimation Using Probe...

Zhao, Yan.

FindBook

Google Book

Amazon

博客來

Traffic State Estimation Using Probe Vehicle Data.

紀錄類型:	書目-電子資源 : Monograph/item
正題名/作者:	Traffic State Estimation Using Probe Vehicle Data./
作者:	Zhao, Yan.
出版者:	Ann Arbor : ProQuest Dissertations & Theses, : 2020,
面頁冊數:	152 p.
附註:	Source: Dissertations Abstracts International, Volume: 81-11, Section: A.
Contained By:	Dissertations Abstracts International81-11A.
標題:	Civil engineering. -
電子資源:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=28006636
ISBN:	9798643184522

Traffic State Estimation Using Probe Vehicle Data.
Zhao, Yan.

Traffic State Estimation Using Probe Vehicle Data. - Ann Arbor : ProQuest Dissertations & Theses, 2020 - 152 p.

Source: Dissertations Abstracts International, Volume: 81-11, Section: A.

Thesis (Ph.D.)--University of Michigan, 2020.

This item must not be sold to any third party vendors.

Traffic problems are becoming a burden on cities across the world. To prevent traffic accidents, mitigate congestion, and reduce fuel consumption, a critical step is to have a good understanding of traffic. Traditionally, traffic conditions are monitored primarily by fixed-location sensors. However, fixed-location sensors only provide information about specific locations, and the installation and maintenance cost is very high. The advances in GPS-based technologies, such as connected vehicles and ride-hailing services, provide us an alternative approach to traffic monitoring. While these types of GPS-equipped probe vehicles travel on the road, a vast amount of trajectory data are being collected. As probe vehicle data contain rich information about traffic conditions, they have drawn much attention from both researchers and practitioners in the field of traffic management and control. Extensive literature has studied the estimation of traffic speeds and travel times using probe vehicle data. However, as for queue lengths and traffic volumes, which are critical for traffic signal control and performance measures, most of the existing estimation methods based on probe vehicles can hardly be implemented in practice. The main obstacle is the low market penetration of probe vehicles. Therefore, in this dissertation, we aim to develop probe vehicle based traffic state estimation methods that are suitable for the low penetration rate environment and can potentially be implemented in the real world.First, we treat the traffic state in each location and each time point independently. We focus on estimating the queues forming at isolated intersections under light or moderate traffic. The existing methods often require prior knowledge of the queue length distribution or the probe vehicle penetration rate. However, these parameters are not available beforehand in real life. Therefore, we propose a series of methods to estimate these parameters from historical probe vehicle data. Some of the methods have been validated using real-world probe vehicle data.Second, we study traffic state estimation considering temporal correlations. The correlation of queue lengths in different traffic signal cycles is often ignored by the existing studies, although the phenomenon is commonly-observed in real life, such as the overflow queues induced by oversaturated traffic. To fill the gap, we model such queueing processes and observation processes using a hidden Markov model (HMM). Based on the HMM, we develop two cycle-by-cycle queue length estimation methods and an algorithm that can estimate the parameters of the HMM from historical probe vehicle data.Lastly, we consider the spatiotemporal correlations of traffic states, with a focus on the estimation of traffic volumes. With limited probe vehicle data, it is difficult to estimate traffic volumes accurately if we treat each location and each time slot independently. Noticing that traffic volumes in different locations and different time slots are correlated, we propose to find the low-rank representation of traffic volumes and then reconstruct the unknown values by fusing probe vehicle data and fixed-location sensor data. Test results show that the proposed methods can reconstruct the traffic volumes accurately, and they have great potential for real-world applications.In summary, this thesis systematically studies traffic state estimation based on probe vehicle data. Some of the proposed methods have been implemented in real life. We expect the methods to be implemented on an even larger scale and help transportation agencies solve more real-world traffic problems.

ISBN: 9798643184522Subjects--Topical Terms:

860360
Civil engineering.
Subjects--Index Terms:

Traffic state estimation

Traffic State Estimation Using Probe Vehicle Data.
LDR:04855nmm a2200373 4500 001 2271463
005 20201007134730.5
008 220629s2020 ||||||||||||||||| ||eng d
020 $a 9798643184522
035 $a (MiAaPQ)AAI28006636
035 $a (MiAaPQ)umichrackham003064
035 $a AAI28006636
040 $a MiAaPQ $c MiAaPQ
100 1 $a Zhao, Yan. $3 1033755
245 1 0 $a Traffic State Estimation Using Probe Vehicle Data.
260 1 $a Ann Arbor : $b ProQuest Dissertations & Theses, $c 2020
300 $a 152 p.
500 $a Source: Dissertations Abstracts International, Volume: 81-11, Section: A.
500 $a Advisor: Liu, Henry;Peng, Huei.
502 $a Thesis (Ph.D.)--University of Michigan, 2020.
506 $a This item must not be sold to any third party vendors.
506 $a This item must not be added to any third party search indexes.
520 $a Traffic problems are becoming a burden on cities across the world. To prevent traffic accidents, mitigate congestion, and reduce fuel consumption, a critical step is to have a good understanding of traffic. Traditionally, traffic conditions are monitored primarily by fixed-location sensors. However, fixed-location sensors only provide information about specific locations, and the installation and maintenance cost is very high. The advances in GPS-based technologies, such as connected vehicles and ride-hailing services, provide us an alternative approach to traffic monitoring. While these types of GPS-equipped probe vehicles travel on the road, a vast amount of trajectory data are being collected. As probe vehicle data contain rich information about traffic conditions, they have drawn much attention from both researchers and practitioners in the field of traffic management and control. Extensive literature has studied the estimation of traffic speeds and travel times using probe vehicle data. However, as for queue lengths and traffic volumes, which are critical for traffic signal control and performance measures, most of the existing estimation methods based on probe vehicles can hardly be implemented in practice. The main obstacle is the low market penetration of probe vehicles. Therefore, in this dissertation, we aim to develop probe vehicle based traffic state estimation methods that are suitable for the low penetration rate environment and can potentially be implemented in the real world.First, we treat the traffic state in each location and each time point independently. We focus on estimating the queues forming at isolated intersections under light or moderate traffic. The existing methods often require prior knowledge of the queue length distribution or the probe vehicle penetration rate. However, these parameters are not available beforehand in real life. Therefore, we propose a series of methods to estimate these parameters from historical probe vehicle data. Some of the methods have been validated using real-world probe vehicle data.Second, we study traffic state estimation considering temporal correlations. The correlation of queue lengths in different traffic signal cycles is often ignored by the existing studies, although the phenomenon is commonly-observed in real life, such as the overflow queues induced by oversaturated traffic. To fill the gap, we model such queueing processes and observation processes using a hidden Markov model (HMM). Based on the HMM, we develop two cycle-by-cycle queue length estimation methods and an algorithm that can estimate the parameters of the HMM from historical probe vehicle data.Lastly, we consider the spatiotemporal correlations of traffic states, with a focus on the estimation of traffic volumes. With limited probe vehicle data, it is difficult to estimate traffic volumes accurately if we treat each location and each time slot independently. Noticing that traffic volumes in different locations and different time slots are correlated, we propose to find the low-rank representation of traffic volumes and then reconstruct the unknown values by fusing probe vehicle data and fixed-location sensor data. Test results show that the proposed methods can reconstruct the traffic volumes accurately, and they have great potential for real-world applications.In summary, this thesis systematically studies traffic state estimation based on probe vehicle data. Some of the proposed methods have been implemented in real life. We expect the methods to be implemented on an even larger scale and help transportation agencies solve more real-world traffic problems.
590 $a School code: 0127.
650 4 $a Civil engineering. $3 860360
650 4 $a Transportation. $3 555912
650 4 $a Mechanical engineering. $3 649730
653 $a Traffic state estimation
653 $a Probe vehicle
653 $a Data mining
690 $a 0548
690 $a 0709
690 $a 0543
710 2 $a University of Michigan. $b Mechanical Engineering. $3 2104815
773 0 $t Dissertations Abstracts International $g 81-11A.
790 $a 0127
791 $a Ph.D.
792 $a 2020
793 $a English
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=28006636