Author: Jun Liu
Publisher:
Published: 2015
Total Pages: 173
ISBN-13:
DOWNLOAD EBOOK →
Increasing amounts of data, generated by electronic sensors from various sources that include travelers, vehicles, infrastructure and the environment, referred to as “Big Data”, represent an opportunity for innovation in transportation systems and toward achieving safety, mobility and sustainability goals. The dissertation takes advantage of large-scale trajectory data coupled with travel behavioral information and containing 78 million second-by-second driving records from 100 thousand trips made by nearly four thousand drivers. The data covers 70 counties across the State of California and Georgia, representing various land use types, roadway network conditions and population. The trajectories cover various driving practices made by vehicles of varied body types as well as different fuel types including conventional vehicles (CVs) consuming gasoline, hybrid electric vehicles (HEVs), battery electric vehicles (BEVs), diesel vehicles and other alternative fuel vehicles (AFVs). The dissertation establishes a framework for the research agenda in instantaneous driving behavior studies using the large-scale trajectory data. The dissertation makes both theoretical and empirical contributions: 1) Developing measures for driving volatility in instantaneous driving behaviors; 2) Understanding correlates of driving volatility in hierarchies & developing applications using large-scale trajectory data. Before using second-by-second trajectories, a study, answering research questions concerning the relationships between data sampling rates and information loss, was conducted. Then, a study for quantifying driving volatility in instantaneous driving behaviors was presented. “Driving volatility”, as the core concept in the dissertation, captures extreme driving patterns under seemingly normal conditions. After that, the dissertation presents a study on exploration of the hierarchical nature of driving volatility embedded in travel survey data using multi-level modeling techniques, and highlights the role of AFVs in travel. Last, the dissertation presents a study for customizing driving cycles for individuals using large-scale trajectory data, given heterogeneous driving performance across drivers and vehicle types. The customized driving cycles help generate more accurate fuel economy information to support cost-effective vehicle choices. The implications of the findings and potential applications to fleet vehicles and driving population are also discussed in the dissertation.