Abstract

Disruption is a serious and common problem for the airline industry. High utilisation of aircraft and airport resources mean that disruptive events can have large knock-on effects for the rest of the schedule. The airline must rearrange their schedule to reduce the impact. The focus in this paper is on the Aircraft Recovery Problem. The complexity and uncertainty involved in the industry makes this a difficult problem to solve. Many deterministic modelling approaches have been proposed, but these struggle to handle the inherent variability in the problem. This paper proposes a multi-fidelity modelling framework, enabling uncertain elements of the environment to be included within the decision making process. We combine a deterministic integer program to find initial solutions and a novel simulation optimisation procedure to improve these solutions. This allows the solutions to be evaluated whilst accounting for the uncertainty of the problem. The empirical evaluation suggests that the combination consistently finds good rescheduling options.

摘要

中断是航空业面临的一个严重而普遍的问题。飞机和机场资源的高利用率意味着破坏性事件会对其余的时间表产生巨大的连锁反应。航空公司必须重新安排他们的时间表以减少影响。本文的重点是飞机恢复问题。行业所涉及的复杂性和不确定性，使其成为一个难以解决的问题。已经提出了许多确定性建模方法，但这些方法难以处理问题中固有的可变性。本文提出了一个多保真度建模框架，使环境中的不确定因素能够包含在决策过程中。我们结合了一个确定性整数程序来寻找初始解决方案和一个新颖的模拟优化程序来改进这些解决方案。这允许在考虑问题的不确定性的同时评估解决方案。实证评估表明，该组合始终可以找到良好的重新安排选项。