Ph.D. Proposal

Hyungu Choi

(Advisor: Prof. Dimitri Mavris)

"PREDICTIVE MODELING OF AIRCRAFT ARRIVAL TIMES IN THE TERMINAL MANEUVERING AREA USING DATA-DRIVEN TECHNIQUES"

Tuesday, June 27

12:00 p.m., EDT 

Collaborative Design Environment (CoVE) 

Weber Space Science and Technology Building (SST II) 

And

Click here to join the meeting

Meeting ID: 290 743 376 903

Passcode: wtoFFt

Abstract

This research seeks to solve the rising issue of flight delays within the U.S. aviation industry. The significant problem is that the demand for air travel is growing faster than our ability to manage it effectively. The study focuses on the Terminal Maneuvering Area (TMA), a part of the National Airspace System (NAS) that frequently experiences high traffic and congestion. Instead of suggesting costly and extensive infrastructure expansion, the study proposes better use of the existing air traffic control procedures. One major part of this approach is to improve the accuracy of the Estimated Time of Arrival (ETA) predictions for flights within the TMA. 

In striving towards this goal, the study employs data-driven techniques using flight-related data to improve ETA predictions in the TMA. The proposed methodology comprises three critical steps: 1) Identifying Trajectory Patterns through a Two-Stage Clustering Analysis, 2) Predicting Arrival Time by Incorporating TMA Weather Data, and 3) Predicting Traffic Volume by Integrating Temporal and Spatial Characteristics. This methodology utilizes flight-related data, including ADS-B trajectory and weather data, from aircraft arriving at O'Hare International Airport. 

The first step merges the Gaussian mixture model (GMM) and Density-Based Spatial Clustering of Applications with Noise (DBSCAN) techniques to identify trajectory patterns. This hybrid approach considers both factors influencing trajectory changes and density-based characteristics. Subsequently, the arrival time prediction step seeks to refine the prediction model's accuracy within the TMA by incorporating weather data along the aircraft's path. Weather, especially wind, substantially impacts aircraft operations throughout the flight, from takeoff to landing. Hence, incorporating weather data into the prediction model is imperative to enhance its predictive performance. Finally, recognizing the interconnectivity within the aviation network, the traffic volume prediction step enhances prediction accuracy in the TMA by integrating temporal and spatial characteristics. This improvement is accomplished by combining Temporal Convolutional Networks (TCN) with Graph Convolutional Networks (GCN). 

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