語系:
繁體中文
English
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
Learning in Capacitated Multimodal N...
~
Xu, Jia.
FindBook
Google Book
Amazon
博客來
Learning in Capacitated Multimodal Networks Over Time.
紀錄類型:
書目-電子資源 : Monograph/item
正題名/作者:
Learning in Capacitated Multimodal Networks Over Time./
作者:
Xu, Jia.
出版者:
Ann Arbor : ProQuest Dissertations & Theses, : 2019,
面頁冊數:
174 p.
附註:
Source: Masters Abstracts International, Volume: 80-08.
Contained By:
Masters Abstracts International80-08.
標題:
Transportation. -
電子資源:
http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=13426700
ISBN:
9780438815230
Learning in Capacitated Multimodal Networks Over Time.
Xu, Jia.
Learning in Capacitated Multimodal Networks Over Time.
- Ann Arbor : ProQuest Dissertations & Theses, 2019 - 174 p.
Source: Masters Abstracts International, Volume: 80-08.
Thesis (Ph.D.)--New York University Tandon School of Engineering, 2019.
This item is not available from ProQuest Dissertations & Theses.
Transportation policies and technologies tend to be very model-dependent because field tests in public infrastructure can have significant risks. Such models, however, have historically focused only on the travelers' behavior as they interacted with roadways or fixed route transit systems. These models are not able to deal with emerging shared mobility systems that include the behavior/decision-making of the mobility providers/operators. Without such models, many field pilots are operating blindly without model-based decision support, resulting in risky failures due to high operating costs or low ridership that cannot be properly explained. Methods developed in this study overcome this problem. As a direct result, public agencies will be better prepared to evaluate shared mobility systems scenarios, explain potential impacts, and guide design to be more successful by accounting for both traveler and operator incentives. Our society is challenged with the concept of sustainable development. A range of new technological advancements is creating opportunities to gather and disseminate unprecedentedly detailed and concurrent information about the transport system. These new challenges require new planning and management strategies in the area of transportation planning. The first task in this study is about learning and inference. A new inference framework is proposed to show how to use data obtained from individual traveler decisions (such as vehicle GPS data) to infer parameters of the network that they traverse, such as travel times or the effects of congestion. For example, machine learning techniques may be able to predict a change in travel patterns in a network, but it is hard to be precise enough to attribute those changes in travel patterns to specific links. Network estimation techniques may be able to quantify this effect but encounter a lot of noise due to the need to make an extra step of estimating population variables before being able to estimate link parameters. The proposed network learning approach, based on multi-agent inverse optimization, makes use of the information from the route choice made by each agent from sample data alone to quantify the effects from each link in the network. This discovery has many applications. It can reduce the computational load for city-level traffic monitoring for city agencies. It can be used to calibrate decision-support models for managing multimodal networks (like bike-share systems, feeder services, or park-and-ride infrastructure). It provides a way to apply more advanced route choice models in networks that can account for diverse population that impose various social/congestion externalities on each other. These are important for city monitoring systems and for dynamic operations where real time network route choice data coming from multiple agents is available. System evaluation is another significant task in this study. When new mobility systems are introduced to a city, how city agencies manage the updated transport system successfully? For example, the sustainability of bikesharing systems depends on the bicycle network connectivity and accessibility. A study of bike lane infrastructure investment impacts on bikesharing ridership is conducted for answering such question "for a city agency, how much does the funding level for bike lanes impact shared bike ridership in a city?" The longitudinal study estimates the relationship between Citi Bike (the bikesharing system in New York City) average daily trip counts and the total length of bike lanes in New York City. The time series analysis can provide new insights into system-level causality and temporal lag characteristics. The challenges with the last task of inference for shared mobility systems involves having a behavioral route choice model that can be used to learn the spatial interactions. To accomplish that purpose, a new route choice model is investigated so that when applied to a shared mobility system I can estimate the parameters pertaining to the capacities in the system. This approach involves a random utility model where one of the parameters relates to capacity effects and this is dependent on flows from other travelers in the network. The model is test out with Citi Bike in New York City. The findings in this dissertation demonstrate innovative solutions to make use of data and information in managing shared mobility systems. Contributions will enhance the strategic toolbox available to public agencies that operate the transportation network, and also the private transportation services that are operated in the network. For future studies, the multi-agent IO learning method will be integrated using multiple sensor sources for online learning system. The longitudinal analysis will be conducted in a more complex multimodal system (e.g. bike and ride) to show a better understanding of infrastructure investment impact which will lead to better informed decision-making process. Other extensions of this research include conducting analysis, such as longitudinal studies and behavior route choice models, to the more complex multimodal systems.
ISBN: 9780438815230Subjects--Topical Terms:
555912
Transportation.
Learning in Capacitated Multimodal Networks Over Time.
LDR
:06301nmm a2200337 4500
001
2262975
005
20191121113952.5
008
220629s2019 ||||||||||||||||| ||eng d
020
$a
9780438815230
035
$a
(MiAaPQ)AAI13426700
035
$a
(MiAaPQ)poly:10431
035
$a
AAI13426700
040
$a
MiAaPQ
$c
MiAaPQ
100
1
$a
Xu, Jia.
$3
1677394
245
1 0
$a
Learning in Capacitated Multimodal Networks Over Time.
260
1
$a
Ann Arbor :
$b
ProQuest Dissertations & Theses,
$c
2019
300
$a
174 p.
500
$a
Source: Masters Abstracts International, Volume: 80-08.
500
$a
Publisher info.: Dissertation/Thesis.
500
$a
Advisor: Chow, Joseph Y.J.
502
$a
Thesis (Ph.D.)--New York University Tandon School of Engineering, 2019.
506
$a
This item is not available from ProQuest Dissertations & Theses.
506
$a
This item must not be sold to any third party vendors.
520
$a
Transportation policies and technologies tend to be very model-dependent because field tests in public infrastructure can have significant risks. Such models, however, have historically focused only on the travelers' behavior as they interacted with roadways or fixed route transit systems. These models are not able to deal with emerging shared mobility systems that include the behavior/decision-making of the mobility providers/operators. Without such models, many field pilots are operating blindly without model-based decision support, resulting in risky failures due to high operating costs or low ridership that cannot be properly explained. Methods developed in this study overcome this problem. As a direct result, public agencies will be better prepared to evaluate shared mobility systems scenarios, explain potential impacts, and guide design to be more successful by accounting for both traveler and operator incentives. Our society is challenged with the concept of sustainable development. A range of new technological advancements is creating opportunities to gather and disseminate unprecedentedly detailed and concurrent information about the transport system. These new challenges require new planning and management strategies in the area of transportation planning. The first task in this study is about learning and inference. A new inference framework is proposed to show how to use data obtained from individual traveler decisions (such as vehicle GPS data) to infer parameters of the network that they traverse, such as travel times or the effects of congestion. For example, machine learning techniques may be able to predict a change in travel patterns in a network, but it is hard to be precise enough to attribute those changes in travel patterns to specific links. Network estimation techniques may be able to quantify this effect but encounter a lot of noise due to the need to make an extra step of estimating population variables before being able to estimate link parameters. The proposed network learning approach, based on multi-agent inverse optimization, makes use of the information from the route choice made by each agent from sample data alone to quantify the effects from each link in the network. This discovery has many applications. It can reduce the computational load for city-level traffic monitoring for city agencies. It can be used to calibrate decision-support models for managing multimodal networks (like bike-share systems, feeder services, or park-and-ride infrastructure). It provides a way to apply more advanced route choice models in networks that can account for diverse population that impose various social/congestion externalities on each other. These are important for city monitoring systems and for dynamic operations where real time network route choice data coming from multiple agents is available. System evaluation is another significant task in this study. When new mobility systems are introduced to a city, how city agencies manage the updated transport system successfully? For example, the sustainability of bikesharing systems depends on the bicycle network connectivity and accessibility. A study of bike lane infrastructure investment impacts on bikesharing ridership is conducted for answering such question "for a city agency, how much does the funding level for bike lanes impact shared bike ridership in a city?" The longitudinal study estimates the relationship between Citi Bike (the bikesharing system in New York City) average daily trip counts and the total length of bike lanes in New York City. The time series analysis can provide new insights into system-level causality and temporal lag characteristics. The challenges with the last task of inference for shared mobility systems involves having a behavioral route choice model that can be used to learn the spatial interactions. To accomplish that purpose, a new route choice model is investigated so that when applied to a shared mobility system I can estimate the parameters pertaining to the capacities in the system. This approach involves a random utility model where one of the parameters relates to capacity effects and this is dependent on flows from other travelers in the network. The model is test out with Citi Bike in New York City. The findings in this dissertation demonstrate innovative solutions to make use of data and information in managing shared mobility systems. Contributions will enhance the strategic toolbox available to public agencies that operate the transportation network, and also the private transportation services that are operated in the network. For future studies, the multi-agent IO learning method will be integrated using multiple sensor sources for online learning system. The longitudinal analysis will be conducted in a more complex multimodal system (e.g. bike and ride) to show a better understanding of infrastructure investment impact which will lead to better informed decision-making process. Other extensions of this research include conducting analysis, such as longitudinal studies and behavior route choice models, to the more complex multimodal systems.
590
$a
School code: 1988.
650
4
$a
Transportation.
$3
555912
650
4
$a
Urban planning.
$3
2122922
690
$a
0709
690
$a
0999
710
2
$a
New York University Tandon School of Engineering.
$b
Civil Engineering.
$3
3350166
773
0
$t
Masters Abstracts International
$g
80-08.
790
$a
1988
791
$a
Ph.D.
792
$a
2019
793
$a
English
856
4 0
$u
http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=13426700
筆 0 讀者評論
館藏地:
全部
電子資源
出版年:
卷號:
館藏
1 筆 • 頁數 1 •
1
條碼號
典藏地名稱
館藏流通類別
資料類型
索書號
使用類型
借閱狀態
預約狀態
備註欄
附件
W9415209
電子資源
11.線上閱覽_V
電子書
EB
一般使用(Normal)
在架
0
1 筆 • 頁數 1 •
1
多媒體
評論
新增評論
分享你的心得
Export
取書館
處理中
...
變更密碼
登入