Research On Experiential Particle Swarm Optimization For Dynamic Aircraft Landing Scheduling

Aircraft landing scheduling(ALS) is one of the core content in the terminal area air traffic flow management(ATFM). ALS ensure the aircraft in ATFM land safely and orderly by assigning them feasible landing time and runway. Through the optimization of ALS can improve the aircrafts' benefit to ensure flight safety under. Now our country solve the ALS mainly by first come first serve method, which has become a major constraining factor in improving terminal area flight scheduling efficiency. So it has been important to enhance the terminal area flight scheduling automation level.Unique encoding and decoding methods for particle swarm optimization(PSO) algorithm are designed based on the ALS's mixed integer programming model. During the encoding, aircraft's landing time which is discrete in the mixed integer programming model is mapped into continuous variables through its continuous property.Then, the PSO can find the answer in continuous "space", which not only make full use of the PSO to solve continuous problems, but also reduce the complexity of the problem compared to other discrete method. During the decoding, the solution will be transformed into discrete form of reality and the safety interval constraint is added to the aircraft's landing time to avoid the infeasible solutions. Then, the experiential of Particle Swarm(EPSO) algorithm is presented by designing ALS's local optimization method which can be particle's experiences. The EPSO's effect is better than PSO by adding the particle experiences in the optimization process of PSO. The experiments in the public data sets OR-Library show that the EPSO's performance is superior to the existing algorithms when on large-scale static data sets.For the imperfect of static model of ALS in dynamic practical, the dynamic model more closed to the practical application is established on the basis of the static model. The dynamic model is more suitable for solve the aircraft flow problem in real environment. And the experience of EPSO to solve the dynamic ALS problem becomes better through the combination between dynamic model and the Receding Horizon Control strategy when considering the problems of practical constraints as far as possible. Finally, advantages on the dynamic model are verified through the simulation in data sets relating OR-Library.