Novel Approaches to Model Congestion Evolution and Dependencies in Complex Road Networks
Abstract
Road network defines a basic template that strongly constrains the development of other urban infrastructure systems, thus plays a prominent role in human mobility and activity analysis, as well as more aggregated level studies that look into the land use and urban growth patterns. One of the most important phenomena on urban road networks is the traffic congestion. It is a typical functional failure process on road networks caused by overload of traffic and poses huge impacts on urban systems. The structural characteristics and functional features of road networks interact in complex ways that jointly determine how and where the congestion emerges, how it propagates, and why the failure patterns look distinct in different networks. Although there is extensive literature on analyzing traffic congestion using conventional flow-based approaches, there are gaps in our knowledge regarding the underlying mechanisms of congestion evolution on road networks. The goal of this dissertation is to investigate the coupling of structure and function of complex road networks, and develop a complete solution to monitor, model and suppress traffic congestion on urban road networks using both data-driven and complex network approaches. Specifically, the dissertation consists of three parts. The first part is devoted to developing new data-driven models and tools to monitor and infer various functional state measures of road networks. These include: (1) a network-wide short-term link travel time estimation model using large-scale taxi trip data without detailed trajectory information; (2) a real-time link-level queue length estimation model using license plate recognition (LPR) data; and (3) a network-wide traffic state estimation and prediction model using partially observed LPR data. The second part of the dissertation studies the underlying mechanisms of the emergence, spreading and recovery of traffic congestion. A novel vertex split-recovery model is proposed to explain and model the congestion evolution process as an equivalent structural process (split and recovery on dual vertices) that embedded within the road network structure. As an extension, a design problem is also formulated based on the vertex split-recovery model, which focuses on suppressing the network-wide traffic congestion level. The last part of this dissertation focuses on the spatial dependency between urban sprawl and road networks. The analysis helps reveal the role of road networks in determining future urban development.
Degree
Ph.D.
Advisors
Ukkusuri, Purdue University.
Subject Area
Transportation|Computer science
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.