Research On Real-time Optimization Method For Aircraft Landing Scheduling

As one of the core contents of Air Traffic Flow Management (ATFM) in terminal area, Aircraft Landing Scheduling (ALS) determines an efficient landing scheduling scheme for a given set of aircrafts in order to insure the safety of aircrafts. Researches about ALS problem have great significance for flight safety and improving flight benefit.ALS is a typical combinatorial optimization problem, the existence of such complex multi-constraint makes this problem an intractable problem; and the real-time requirement increases the difficulty of the solving this problem. When there are too many planes waiting for landing, FCFS may not provide a feasible solution. Until now, there have been two kinds of scheduling algorithms for addressing ALS problem, one is Linear Programming (LP) algorithm, and the other is Computational Intelligence (CI) algorithm. LP shows high effectiveness and correctness, however it can hardly find the global optimal solution owing to lack of the ability for global search. Nevertheless CI not only has the powerful ability for global search, but also is able to handle nonlinear constraint and goal function. Therefore it has become a hot topic to resolve ALS using CI. However CI likely produces large computational cost especially at busy airports. Thereby CI needs to combine some heuristic methods to address ALS better.In this paper, we propose a novel approach to address the ALS problem. Compared to previous studies, we put more emphasis in meeting the requirement of real-time scheduling rather than securing a global optimal solution. Our approach, named Cellular-Automata-based Optimization (CAO) consists of two main steps. First, we aim to seek a considerably good aircraft landing sequence via a fast heuristic search base on a Cellular Automata (CA) model. Second, the landing sequence obtained by the CA model is further fine-tuned by a simple yet effective local search procedure.Experimental study on the standard database was conducted to compare the CAO method and several popular approaches in the literature. It was observed that the CAO method not only managed to attain solutions of higher quality, but also was much faster (in CPU time) than the compared methods. The latter property is of significant importance to applying our method in real-world.