Optimization of Aircraft Maintenance Routing and Improved Robustness Based on an Adaptive Reinforcement Learning Algorithm

In this paper, we investigate optimization of an aircraft maintenance routing problem (AMRP), which is currently a hot research topic in the air transport industry. The optimization of AMRP can help airline companies improve their management capacity and operational efficiency. However, there are several complexities in solving the problem, including that aircraft maintenance costs cannot be taken into account in the execution of flight schedules, and the possibility of flight delays may be ignored when assigning flight routes to aircraft. To address these challenges, we first describe Model I by considering multiple maintenance requirements for AMRP, with the intention of maximizing the profits between the flight connection revenues and the aircraft maintenance costs. Second, we further construct a robust model (Model II) based on the average civil aviation delay time, this should reduce the risk of delays while ensuring the flight connection benefits. Finally, we delve into the field of solution methods and we design an adaptive reinforcement learning-based algorithm for solving Model I and Model II. To prove the validity of our model and algorithm, extensive numerical experiments are conducted. The experimental results demonstrate that the proposed models and algorithm have advantages in terms of speed and solution quality, and we can apply our approach to addressing AMRP in multiple environments.