Λεπτομέρειες βιβλιογραφικής εγγραφής
| Τίτλος: |
Large-Scale Airline Crew Recovery Using Mixed-Integer Optimization and Supervised Machine Learning. |
| Συγγραφείς: |
Hızır, Ahmet Esat1 (AUTHOR) aehizir@mit.edu, Barnhart, Cynthia2 (AUTHOR) cbarnhar@mit.edu, Vaze, Vikrant3 (AUTHOR) vikrant.s.vaze@dartmouth.edu |
| Πηγή: |
Transportation Science (INFORMS). Mar/Apr2026, Vol. 60 Issue 2, p177-196. 20p. |
| Θεματικοί όροι: |
Mixed integer linear programming, Supervised learning, Combinatorial optimization, Flight delays & cancellations (Airlines) |
| Περίληψη: |
Airlines take a variety of actions to recover schedules of their aircraft, crew, and passengers from operational disruptions. Aircraft are typically recovered first, followed by crew recovery, and then passenger recovery. This paper aims to repair disrupted crew schedules while ensuring the feasibility of previously decided aircraft recovery plans and indirectly accounting for passenger disruption costs. We develop a fast solution approach that effectively combines mixed-integer optimization and supervised machine learning (ML) methods to find high-quality solutions to large-scale recovery problems. Our approach reduces the solution space by adding constraints based on the patterns discovered in the solutions to offline instances. The model with the added constraints is solved using a mixed-integer optimization solver. To account for the fact that the available time for airlines to handle disruptions may vary during the day of operations, our solution approach allows parameter tuning to flexibly match the extent of solution space reduction to the available runtime. This helps the proposed method to effectively navigate the trade-off between solution quality and runtime. Extensive computational experiments with actual flight and crew schedules of a major U.S. airline with more than 2,800 daily flights show that our approach consistently generates solutions of significantly higher quality than benchmarks and is estimated to provide tens of millions of dollars of reduction in annual operating costs. Moreover, our ML models have interpretable structures that are critical to enhance end-user trust in the ML recommendations. Finally, our approach yields solutions that are more robust to uncertainty in delay prediction than those found by direct optimization. Supplemental Material: The online appendix is available at https://doi.org/10.1287/trsc.2025.0105. [ABSTRACT FROM AUTHOR] |
|
Copyright of Transportation Science (INFORMS) is the property of INFORMS: Institute for Operations Research & the Management Sciences and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.) |
| Βάση Δεδομένων: |
Business Source Index |