Monitoring and control of transport networks using parsimonious models

Mat Jusoh, Ruzanna (2019) Monitoring and control of transport networks using parsimonious models. PhD thesis, University of Glasgow.

Full text available as:
[thumbnail of 2019MatJusohPhD.pdf] PDF
Download (12MB)
Printed Thesis Information: https://eleanor.lib.gla.ac.uk/record=b3348538

Abstract

The growing number of vehicles on the roads coupled with inefficient road operations have generated traffic congestion. Consequently, traffic congestion increase trip time and indirectly contributes to poor quality of life and environmental pollution. Therefore, alleviating traffic congestion, especially in urban networks, is crucial and requires efficient traffic management and control. Recently, macroscopic operational scheme has become the preferred method for monitoring and mitigating traffic congestion due its simplicity in modeling complex large-scale cities and low computational effort. The schemes are based on parsimonious models known as Macroscopic or Network Fundamental Diagram (MFD or NFD) which provides an aggregated model of urban traffic dynamics, linking network circulating flow and average density. This thesis deals with an open problems associated with two main applications of NFD in transportation networks, namely: 1) Traffic monitoring and 2) Traffic flow control. Two parts of the thesis concentrates on each application separately.

The implementation of NFD in perimeter control strategy requires an accurate estimation of NFD where its measurements are reflected from sensors located at appropriate locations in the network. First part of the thesis elaborates a new approach for studying sensor selection for the development of operational or sparse-measurement NFD, with less number of sensor and associated measurements. An information-theoretic based framework is proposed for the optimal sensor selection across a transport network to assist an efficient model selection and construction of sparse-measurement NFD. For the optimal sensor selection, a generalised set covering integer programming (GIP) is developed. Under this framework, several tools to assess GIP solutions are uitilised. First, a correlation between variables is introduced as a ''distance'' metric rather than spatial distance to provide sufficient coverage and information accuracy. Second, the optimal cost of GIP problem is used to determine minimum number of sensors. Third, the relative entropy or Kullback-Leibler divergence is used to measure the dissimilarity between probability mass functions corresponding to different solutions of the GIP program. The proposed framework is evaluated with experimental loop-detector data of one week from central business district with fifty-eight sensors. Results reveal that the obtained sparse-measurement diagrams from the selected models adequately preserve the shape and the main features similar to a full-measurement diagram. Specifically, the coverage level of 24% of the network demonstrated the effectiveness of GIP framework. Simulation results also disclose the Kullback-Liebler divergence as more generic and reliable metric of information loss. Such framework can be of great importance towards a cost-effective sensors installation and maintenance whilst improving the estimation of NFD for better monitoring and control strategy.

Second part of the thesis discusses the traffic flow control problem involving single input flow distribution from perimeter control strategy towards number of gated links at the periphery of the network. It if often assumed that input flow ordered by perimeter control strategy should be equally distributed to a number of candidate junctions. There has not been considerable research into limited storage capacity/different geometric characteristics at gated links as well as equity properties for driver. A control scheme for the multi-gated perimeter flow control (MGC) problem is developed. The scheme determines optimally distributed input flows for a number of gates located at the periphery of a protected network area. A parsimonious model is employed to describe the traffic dynamics of the protected network. To describe traffic dynamics outside of the protected area, the basic state space model is augmented with additional state variables for the queues at store-and-forward origin links of the periphery. The perimeter traffic flow control problem is formulated as a convex optimal control problem with constrained control and state variables. For the application of the proposed scheme in real time, the optimal control problem may be embedded in a rolling-horizon scheme using the current state of the whole system as the initial state as well as predicted demand flows at origin/entrance links. This part also offers flow allocation policies for single-region network without considering entrance link dynamics namely capacity-based flow allocation policy and optimisation-based flow allocation policy. Simulation results are carried for a protected network area of downtown San Francisco with fifteen gates of different geometric characteristics. Results demonstrate the efficiency and equity properties of the MGC approach to better manage excessive queues outside of the protected network area and optimally distribute the input flows. The MGC outperforms the other approaches in terms of serving more trips in protected network as well as shorter queues at gated links. Such framework is particularly of interest to city managers because the optimal flow distribution may influence the network throughput hence serves maximum number of network users.

Item Type: Thesis (PhD)
Qualification Level: Doctoral
Keywords: Network fundamental diagram, multi-gated perimeter flow control, near optimal sensor selection, information theory.
Colleges/Schools: College of Science and Engineering > School of Engineering > Infrastructure and Environment
Supervisor's Name: Ampountolas, Dr. Konstantinos
Date of Award: 2019
Depositing User: Mrs Ruzanna Mat Jusoh
Unique ID: glathesis:2019-41160
Copyright: Copyright of this thesis is held by the author.
Date Deposited: 25 Apr 2019 09:37
Last Modified: 05 Mar 2020 21:47
Thesis DOI: 10.5525/gla.thesis.41160
URI: https://theses.gla.ac.uk/id/eprint/41160
Related URLs:

Actions (login required)

View Item View Item

Downloads

Downloads per month over past year