Research On Modeling Method Of Flight Support Service For Hub Airport

By studying the flight support service process,accurate forecasting of flight support service time is achieved,thereby improving the normal rate of flights.The research content in this paper is mainly divided into two parts.One part is to establish the corresponding mathematical model of flight support service on the basis of analyzing the support service process;the other part is to realize the accurate prediction of flight support service time based on the established model.The research on flight support service process is a very complicated nonlinear problem.Through the analysis of the actual process,a BP neural network-based modeling method is proposed.The number of hidden layer neurons in the network model and the connection weight between the layers of the network are optimized.The incentive function of the model is determined by the progressive function comparison method to establish the flight support service neural network model.On the one hand,based on the network model,an improved genetic algorithm is introduced to optimize the model.The improvement of genetic algorithm is mainly reflected in chromosome structure,fitness function,selection,crossover,mutation operator and cross mutation probability.An adaptive multi-layer genetic algorithm BP neural network(AMGA-BPNN)method is proposed to realize the static prediction of flight support service time.On the other hand,for the problem that the BPNN algorithm converges slowly,the Extreme Learning Machine(ELM)is used for optimization.Aiming at the dynamic and nonlinear problems of the flight support service process,the self-feedback layer and local weighting method are introduced to improve,which not only simplifies the global complexity,but also improves the prediction accuracy.A dynamic recursive local weighted extreme learning machine(DRLW-ELM)method is proposed to achieve dynamic prediction of flight support service time.Using the actual flight support service data of a hub airport,the model method is verified,and the prediction results of different methods are compared.The model evaluation results show that the DRLW-ELM algorithm has the highest prediction accuracy for flight support service time.