Full metadata record
| DC Field | Value | Language |
|---|---|---|
| dc.contributor | Department of Civil and Environmental Engineering | en_US |
| dc.contributor.advisor | Lam, H. K. William (CEE) | en_US |
| dc.contributor.advisor | Sumalee, Agachai (CEE) | en_US |
| dc.creator | Fu, Hao | - |
| dc.identifier.uri | https://theses.lib.polyu.edu.hk/handle/200/12157 | - |
| dc.language | English | en_US |
| dc.publisher | Hong Kong Polytechnic University | en_US |
| dc.rights | All rights reserved | en_US |
| dc.title | Sensor location problems for estimation of origin-destination demands and travel times under uncertainty | en_US |
| dcterms.abstract | To monitor traffic congestion and improve road network performance, various types of traffic sensors have become available and affordable with the rapid development of advanced sensing technologies. Smartly determining the locations of the multi-type sensors is crucial to collect multi-source data for the development of strategic transport models. In view of this, this thesis proposes a new modeling approach to optimize the number and locations of multi-type traffic sensors by taking into account the traffic demand variation and/or travel time uncertainty. The optimum deployment of multi-type traffic sensors is proficient in the separate and simultaneous estimation of day-to-day vehicular traffic demand by origin-destination (OD) pair and travel time on links with covariance effects on a daily scale. The novelty of the research presented in this thesis mainly resides in the incorporation of covariance of OD demands and/or link travel times when deploying single-type or multi-type traffic sensors onto a road network for updating strategic transport models. | en_US |
| dcterms.abstract | In literature, most of the existing methods estimate only the mean (average) OD demands using the observed data from a single-type sensor system. However, vehicular traffic demands between different OD pairs in a typical hourly period (e.g., morning peak hour) are statistically correlated from day to day because of daily variation in activity patterns. Traffic demands during different hourly periods within a day are also highly interrelated, owing to the hourly variation of travel patterns. Moreover, travel times on road links during the peak hour period are stochastic and correlated, especially the travel times of adjacent links under congested conditions. | en_US |
| dcterms.abstract | To overcome the limitation of the existing methods, both the mean and covariance of OD demand and of link/path travel time will be estimated separately and simultaneously with making use of the various data from different sensor systems. In summary, the following key contributions of the thesis are highlighted. | en_US |
| dcterms.abstract | First, spatial covariance of peak-hour OD demand between different OD pairs is explicitly considered in traffic sensor (i.e., traffic count) location problems. The vehicular traffic demands between different OD pairs in a typical hourly period (e.g., the morning peak hour) can be statistically correlated from day to day because of joint travel behaviors and daily variation in activity patterns over a year. A new criterion based on the weighted maximum possible relative error is employed to measure the estimation accuracy of OD demand covariance (i.e., the maximum estimation error of the worst case) without the need for the ground truth of OD demand covariance. A new model is then developed to optimize traffic sensor (i.e., traffic count) locations and thereby minimize the new criterion. The conventional traffic sensor location model is therefore a special case of this new model. | en_US |
| dcterms.abstract | Second, this traffic sensor location model is extended to optimize the location of multi-type traffic sensors by incorporating the spatiotemporal covariance of vehicular traffic demands between different OD pairs in multiple periods. Due to hourly variation of travel patterns by time of day and day of the year, the traffic demands of OD pairs are highly interrelated during different periods (e.g., morning peak and evening peak hours). Thus, a Kalman filter method based on principal component analysis is developed to estimate multi-period OD demands and their covariances. In addition, a novel model is devised for optimizing the locations of multi-type traffic sensors by minimizing the uncertainty of multi-period OD demand estimates. Overall, both the number and locations of multi-type traffic sensors, including point sensors and automatic vehicle identification (AVI) sensors, are optimized under a constraint on the total available budget. The mathematical properties of the new model are studied to determine the effect of multi-period OD flow covariance on the model results. | en_US |
| dcterms.abstract | Thirdly, to develop consistent strategic transport models, an integrated traffic sensor location model is formulated for simultaneous estimation of OD demands and link travel times with consideration of two sources of spatial covariance. The two sources of spatial covariance include the traffic demand covariance between different OD pairs and the travel time covariance between different links during the peak hour period. Coherent estimations of these stochastic link travel times and OD demands are facilitated by multi-source data from multi-type sensors. With these simultaneous estimations, a multi-type sensor location model is developed to efficiently use or fuse these multi-source data. Based on the data observed from the installed multi-type traffic sensors such as link speed/flow and path travel time information, a novel Kullback-Leibler divergence-based model is proposed to achieve the simultaneous estimation. The proposed model can accommodate different probability distributions of OD demands and link travel times under different traffic conditions. | en_US |
| dcterms.abstract | An improved firefly algorithm is developed for efficiently searching for the near-to-global solution, which enables the efficient solution of multi-type sensor location problems that belong to integer programming and are NP-hard. In this improved algorithm, the search strategy is enhanced by taking into account the mean and covariance of OD demand and link travel time. Numerical examples of synthetic and real-world road networks are conducted to illustrate the applications and merits of the proposed sensor location models for separate and simultaneous estimation of OD demands and link travel times with covariance effects. Consequently, the optimal multi-type sensor location schemes can be determined for estimation of day-to-day peak hour vehicular traffic demands by OD pair and/or link travel times. Based on these results, transportation planners and traffic engineers can easily deploy efficient sensor systems to monitor traffic conditions and assess the congestion levels in road networks with uncertainty. | en_US |
| dcterms.extent | xxiv, 204 pages : color illustrations | en_US |
| dcterms.isPartOf | PolyU Electronic Theses | en_US |
| dcterms.issued | 2022 | en_US |
| dcterms.educationalLevel | Ph.D. | en_US |
| dcterms.educationalLevel | All Doctorate | en_US |
| dcterms.LCSH | Traffic engineering -- Data processing | en_US |
| dcterms.LCSH | Detectors | en_US |
| dcterms.LCSH | Electronics in transportation | en_US |
| dcterms.LCSH | Hong Kong Polytechnic University -- Dissertations | en_US |
| dcterms.accessRights | open access | en_US |
Copyright Undertaking
As a bona fide Library user, I declare that:
- I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
- I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
- I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.
By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.
Please use this identifier to cite or link to this item:
https://theses.lib.polyu.edu.hk/handle/200/12157