2026-09-19T16:00:30Zhttps://dspace.mit.edu/server/oai/request

oai:dspace.mit.edu:1721.1/1629992025-10-07T04:13:59Zcom_1721.1_7582com_1721.1_7581col_1721.1_131023

Fan, Chuchu Zhang, Joseph Massachusetts Institute of Technology. Department of Electrical Engineering and Computer Science 2025-10-06T17:38:53Z 2025-10-06T17:38:53Z 2025-05 2025-06-23T14:04:45.311Z https://hdl.handle.net/1721.1/162999 0009-0000-2680-3547 Understanding the interaction between weather and disruptions in complex air transportation network is important to the design and evaluation of preemptive measures and responses taken by air traffic managers. However, the occurrence of disruptive weather events is often rather limited compared to the amount of data available for nominal operations. Additionally, in large-scale systems with many known and unknown confounding factors, it can be difficult to identify the relevance of existing data to different underlying distributions of interest. Furthermore, existing work generally follows a frequentist paradigm in predicting disruptions based on weather, and does not easily lend itself to inferring the causes of disruptions, which can be important both in building models and using them to make predictions, and generate test cases to stress-test proposed design decisions. In this thesis, we develop a hierarchical Bayesian model for air traffic network operations, and investigate methods for learning these models in data-constrained settings, by extend existing work on retrospectively analyzing failures. We also include a guiding case study performed on LaGuardia Airport, in which a generative model is developed for the interaction between weather conditions and airport-level parameters within a single airport, trained on unlabeled historical data, and evaluated by simulating disruptions on historical schedules. M.Eng. Massachusetts Institute of Technology Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0) Copyright retained by author(s) https://creativecommons.org/licenses/by-nc-nd/4.0/ Risk Management in Air Traffic Applications: Data-Driven Modeling, Prediction, and Generation of Realistic Weather Disruptions and Other Unfavorable Conditions Thesis application/pdf Master Master of Engineering in Electrical Engineering and Computer Science Publication unknown unknown <Publication xmlns="https://www.openaire.eu/cerif-profile/1.1/" id="60843785-7be9-4c9d-8f82-5313e97105fb"> <Type xmlns="https://www.openaire.eu/cerif-profile/vocab/COAR_Publication_Types">http://purl.org/coar/resource_type/c_1843</Type> <Title>Risk Management in Air Traffic Applications: Data-Driven Modeling, Prediction, and Generation of Realistic Weather Disruptions and Other Unfavorable Conditions</Title> <PublishedIn> <Publication> </Publication> </PublishedIn> <PublicationDate>2025-05</PublicationDate> <Authors> <Author> <DisplayName>Zhang, Joseph</DisplayName> <Affiliation> <OrgUnit> </OrgUnit> </Affiliation> </Author> </Authors> <Editors> </Editors> <Publishers> <Publisher> <DisplayName>Massachusetts Institute of Technology</DisplayName> <OrgUnit /> </Publisher> </Publishers> <License>https://creativecommons.org/licenses/by-nc-nd/4.0/</License> <Abstract>Understanding the interaction between weather and disruptions in complex air transportation network is important to the design and evaluation of preemptive measures and responses taken by air traffic managers. However, the occurrence of disruptive weather events is often rather limited compared to the amount of data available for nominal operations. Additionally, in large-scale systems with many known and unknown confounding factors, it can be difficult to identify the relevance of existing data to different underlying distributions of interest. Furthermore, existing work generally follows a frequentist paradigm in predicting disruptions based on weather, and does not easily lend itself to inferring the causes of disruptions, which can be important both in building models and using them to make predictions, and generate test cases to stress-test proposed design decisions. In this thesis, we develop a hierarchical Bayesian model for air traffic network operations, and investigate methods for learning these models in data-constrained settings, by extend existing work on retrospectively analyzing failures. We also include a guiding case study performed on LaGuardia Airport, in which a generative model is developed for the interaction between weather conditions and airport-level parameters within a single airport, trained on unlabeled historical data, and evaluated by simulating disruptions on historical schedules.</Abstract> <Access xmlns="http://purl.org/coar/access_right" > </Access> </Publication>