The Research Of Robust Gate Assignment Method

With the rapid development of civil aviation transportation industry,the passenger traffic carried by some of the hub airports has been approaching to its saturation.In the case of not expanding or building new airports,how to utilize the airport gate resources scientifically and reasonably has become the primary problem to be solved at the airport.The gate assignment problem(GAP)aims to assign gates to the aircrafts considering the operational efficiency of the airport and the satisfaction of the passengers.Meanwhile,the impact of unexpected factors,such as bad weather,flight delays or early arrivals should be considered.This requires that our scheduling model should have the ability to handle the dynamic changes and be able to absorb the slight changes in flight time.That is to say,the schedule flexibility is required.The research work in this thesis mainly includes the following points: First,a constrained biobjective robust gate assignment model(CBR-GAM)is proposed in this paper.The minimization of passengers' total walking distances and the total robust cost of the gate assignment are treated as the two objectives.Second,to address the CBR-GAM efficiently,we propose a two-phase large neighborhood search(2PLNS).In the first phase,the bi-objective optimization problem is decomposed into a set of single-objective subproblems using weighted sum approach.One solution is associated with one subproblem and the local search heuristic is conducted on it.The solution set(population)aims to converge fast towards the PF in the first phase.In the second phase,Pareto local search(PLS)is applied to solution sets obtained in the first phase for more Pareto approximated solutions.Third,2PLNS uses a greedy and stochastic strategy(GSS)for the large neighborhood search,to improve the convergence speed and avoid the search being stuck in the local optima.Fourth,test cases based on the actual data of the airport and simulated data are targeted in this thesis.The experimental results show that 2PLNS has good convergence and diversity.In addition,we also discuss how to select several solutions of interest from a large number of nondominated solutions.