A Q-Learning-Based Particle Swarm Optimization for Aircraft Routing and Scheduling in Airport Terminal Area

As the airport terminal area becomes progressively crowded, costly delays and adverse environmental impact due to excessive fuel burn require effective aircraft routing and scheduling in the airport terminal area. Aircraft routing and scheduling in the airport terminal area refers to the safe and efficient movement of aircraft among airport facilities such as runways, aircraft stands and taxiways. It is a hybrid optimization problem that involves both the airport ground movement problem and the aircraft sequencing problem. The paper investigates the hybrid problem with constraints such as the safe distance between aircraft, queuing limit at the runways, and release time difference for sake of safety. The objective is to minimize the average taxiing time of the aircraft. A particle swarm optimization and Q-learning based aircraft routing and scheduling algorithm (PSO-QL-ARS) is proposed for the problem considered. The proposal adopts the main framework of PSO. It consists of three major components: the ideal shortest path algorithm, the semi-no-wait schedule generation method and the Q-learning based particle evolving method. The shortest path algorithm takes into consideration turning time according to the turning angles on the path and different taxiing speeds depending on the taxiway types. The semi-no-wait schedule generation method is presented to compute the feasible trajectory plan for an aircraft, including the waiting time at each vertex and the taxiing time on each taxiway. It attempts to place the waiting time of an aircraft on the aircraft stand. The Q-learning-based particle evolving method employs a Q-table to select the finest evolving action which is used to update the position of the given particle. The proposed algorithm is compared with four baseline algorithms. The experimental results show that the proposal outperforms the compared baseline algorithms in effectiveness and robustness.