Integrated Corridor Management Framework For Severe Freeway Incidents

Author

• Sanjida Afroz Iqra, University of Central FloridaFollow

ORCID

https://orcid.org/0009-0005-6431-9103

Keywords

Integrated corridor management, Traffic incident management, Congestion, Traffic speed forecasting, Graph neural networks, Connected vehicle data

Subject Categories

Civil Engineering | Systems and Communications | Transportation Engineering

Abstract

Traffic incidents are a major source of non-recurrent congestion on urban freeways, generating substantial mobility, safety, and economic impacts. Severe incidents that block multiple or all travel lanes are particularly disruptive because they degrade freeway operations and propagate congestion onto surrounding arterials. Effective Integrated Corridor Management (ICM) requires the ability to identify severe incidents, estimate their network-wide impacts, and anticipate the traffic conditions and driver behaviors that contribute to instability. This dissertation develops a data-driven ICM framework to address these challenges using real-world incident, crash, detector, and connected vehicle data from major Central Florida corridors, including I-4 and SR-417. The first component establishes a framework for classifying and predicting severe post-crash delay. A Bayesian Gaussian Mixture Model identifies congestion patterns using spatiotemporal congestion characteristics and travel-time reliability measures, while machine-learning models forecast delay evolution across successive post-crash intervals to support real-time operational decision making. The second component examines the network-wide impacts of Freeway Crash Induced All-Lane Closure (FCIALC) events. A hierarchical Bayesian Network models the factors associated with full lane closures, and a copula-based Bayesian regression quantifies the dependence between closure severity and arterial congestion. Complementing this macro-level analysis, a Dual-Encoder Graph-Attention Mixture-of-Experts (DE-GAM) architecture predicts post-incident traffic speeds across freeways, arterials, and detour routes by integrating baseline traffic conditions, incident information, and dynamic message sign data through temporal and spatial attention mechanisms. The third component advances proactive traffic management by leveraging connected vehicle trajectories to identify risky drivers and predict their future locations. A multi-modal Transformer captures persistent journey-level risky-driving patterns from long-term driving history, while a graph-attention-based location prediction model forecasts downstream roadway occupancy. Together, these contributions provide an integrated, interpretable, and operationally oriented framework that advances traffic management from reactive incident response toward proactive, network-aware control, supporting more effective congestion mitigation, traveler guidance, and corridor management strategies.

Completion Date

2026

Semester

Summer

Committee Chair

Abdel-Aty, Mohamed

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

Language

English

STARS Citation

Iqra, Sanjida Afroz, "Integrated Corridor Management Framework For Severe Freeway Incidents" (2026). Graduate Studies Theses and Dissertations 2026. 281.
https://stars.library.ucf.edu/gradstudies_etd_2026/281