ORCID

0000-0002-2976-0762

Keywords

Connected Vehicle Data, Traffic Safety, Traffic Mobility, Driving Behavior, Deep Learning

Abstract

Traffic safety and mobility remain critical challenges for modern transportation systems. The emergence of connected vehicle (CV) data offers unprecedented opportunities to capture microscopic, non-aggregated driving dynamics to support precise and finer-grained traffic safety and mobility research. This dissertation develops a comprehensive CV data–driven analytical framework to advance traffic safety and mobility research across multiple roadway contexts, including intersections, segments, freeways, and urban arterial networks. For traffic safety, this research extracts both longitudinal and lateral risky driving behaviors from CV trajectories and integrates them with macro-level roadway, traffic, and visual environment features. A spatial machine learning framework is proposed for intersection crash analysis to capture nonlinear effects and spatial heterogeneity, revealing the critical role of risky turning behaviors. Segment-level safety analysis further incorporates lane-changing behavior and driving volatility using unidirectional analysis units and multivariate statistical models to quantify heterogeneous impacts across crash types. For traffic mobility, a macro–micro cross-attention transformer model is developed to jointly model interactions between aggregated traffic flow states and micro-level driving behaviors for freeway speed prediction, demonstrating improved accuracy under congested conditions and varied forecasting horizons. At the urban arterial network level, a pure CV data–based approach is proposed to estimate key traffic state measures, followed by an abnormality-aware spatiotemporal graph convolutional network to enhance delay and queue prediction under both normal and abnormal traffic conditions. Extensive experiments on real-world freeway and arterial networks validate the effectiveness and robustness of the proposed methods. Overall, this dissertation demonstrates the value of behavior-aware, CV data–driven modeling for improving traffic safety analysis and proactive mobility management, providing scalable tools to support data-driven ITS operations.

Completion Date

2026

Semester

Spring

Committee Chair

Mohamed Abdel-Aty

Degree

Doctor of Philosophy (Ph.D.)

College

College of Engineering and Computer Science

Department

Department of Civil, Environmental and Construction Engineering

Format

PDF

Document Type

Dissertation

Identifier

DP0053093

Download

Share

COinS
 

Accessibility Statement

This item was created or digitized prior to April 24, 2027, or is a reproduction of legacy media created before that date. It is preserved in its original, unmodified state specifically for research, reference, or historical recordkeeping. In accordance with the ADA Title II Final Rule, the University Libraries provides accessible versions of archival materials upon request. To request an accommodation for this item, please submit an accessibility request form.