Hyungu Choi
(Advisor: Prof. Dimitri Mavris] 
will defend a doctoral thesis entitled, 

PREDICTIVE MODELING OF AIRCRAFT ARRIVAL TIMES IN THE TERMINAL MANEUVERING AREA USING DATA-DRIVEN TECHNIQUES 
On 
Wednesday, August 6 at 2:00 p.m., EDT 
Collaborative Visualization Environment (CoVE) 
Weber Space Science and Technology Building (SST II)

And
Join the meeting now
Meeting ID: 250 235 737 252 5
Passcode: ey3Jf9tQ
Abstract

This research addresses the growing challenge of congestion and flight delays in the U.S. aviation system by focusing on operational inefficiencies within the Terminal Maneuvering Area (TMA). As the demand for air travel continues to grow faster than the system’s capacity to manage it, expanding infrastructure alone is no longer a practical or sustainable solution. In response, this study investigates the potential of data-driven methods to improve the predictability of flight arrivals, with particular emphasis on enhancing the accuracy of Estimated Time of Arrival (ETA) predictions within the TMA.
To achieve this objective, the proposed methodology comprises three critical steps: 1) Spatiotemporal Modeling for Airport Traffic Forecasting, 2) Bi-Level Clustering of Arrival Trajectories, and 3) ETA Prediction with Traffic and Weather Features. The approach is applied to flights arriving at Chicago O’Hare International Airport and is based on ADS-B trajectory data and high-resolution weather observations.
The first stage improves short-term traffic forecasting not only at the airport level but also across the network level, by modeling temporal dependencies and spatial interactions among airports using a combination of Long Short-Term Memory (LSTM) networks and Graph Convolutional Networks (GCN). The second stage identifies operationally meaningful trajectory groups through a bi-level clustering framework. This process first applies K-means to categorize trajectories based on operational conditions and then uses DBSCAN to capture variation in path shape through density-based clustering. In the final stage, ETA prediction is refined by incorporating features related to aircraft category, traffic density, and terminal weather conditions, particularly wind, which significantly affects aircraft behavior during all phases of flight. Collectively, this framework supports more accurate arrival predictions in congested airspace and contributes to more efficient air traffic management without requiring significant infrastructure investment.
Committee
•    Prof. Dimitri N. Mavris - School of Aerospace Engineering (Advisor)
•    Prof. Daniel P. Schrage - School of Aerospace Engineering
•    Prof. Polo Chau - School of Computational Science and Engineering
•    Prof. B. Aditya Prakash - School of Computational Science and Engineering
•    Dr. Ameya Behere - Research Engineer, School of Aerospace Engineering