Artificial Intelligence Application for Solving Mixed Integer Linear Programming with Learning Optimization Techniques in Air Transportation and Mobility

• Artur Guerra Rosa
• Felipe Tavares Loureiro
• Marcus Vinicius Santos Da Silva
• Silvia Araujo dos Reis
• Vitor Celestino

University of Brasilia · Pontifical Catholic University of Goias

Evento

XXII SITRAER – Brasília, DF, Brasil, 22 a 24 de outubro de 2025

Volume

Anais do XXII Simpósio de Transporte Aéreo – SITRAER 2025 · ISBN 978-65-272-2024-4 · DOI 10.29327/9786527220244

Editora

Even3 Publicações

Editores

Fabiana Serra de Arruda, Camila Lucena, Giovanna Miceli Ronzani Borille, Evandro José da Silva

Abstract

Solving complex optimization problems in air transportation and mobility can be greatly improved by combining Mixed Integer Linear Programming (MILP) with modern computational techniques. Although MILP provides a solid framework for modeling these challenges, which often involve complex combinatorial structures and discrete decisions, traditional solution methods frequently struggle with the computational demands of real-world scenarios' scale and dynamic nature, largely due to their inherent NP-hard complexity. This narrative literature review highlights how integrating artificial intelligence methods enhances MILP, specifically focusing on Learning to Optimize (L2O) techniques. The L2O approach uses data-driven strategies, learning from historical operational data to improve how optimization algorithms function automatically. Instead of relying on heuristics, these methods can adaptively guide the search for solutions—such as predicting promising pathways or pruning unpromising ones—resulting in quality solutions with known optimality gap. Practical applications of these combined approaches are examined in key areas, such as flight scheduling, airport resource allocation, air traffic management, smart transportation systems, and route optimization. Analysis shows that integrating these methods with MILP can significantly improve operational efficiency, reduce computational time, and support more adaptive and reliable decisionmaking in complex transportation networks. By providing a comprehensive overview of the current state of the art, the research demonstrates the tangible potential of data-informed optimization to advance the transportation sector by overcoming the longstanding limitations of traditional approaches, ultimately fostering more resilient and efficient air transportation and mobility systems.

Keywords: Artificial Intelligence; Mixed Integer Linear Programming; Learning to Optimize; Air Transportation; Air Mobility