Research On Propagation Model And Prediction Of Airport Large-scale Flight Delay

Large-scale flight delays have become one of the tough problems and challenges in the development of civil aviation industry.They not only cause disruption of flight operation,but also may beset the flight schedules of downstream airports.Due to the bad weather,traffic control,military activities and other reasons,the occurrence of large-scale flight delays becomes inevitable.However,it is feasible to predict and control delays and their propagation in advance,which is based on the scientific understanding of the emergence,development and dissipation of large-scale flight delays and their propagation.Aiming at the flight delay propagation problem,two flight delay propagation models are established,and the delay propagation characteristics are concluded.Finally,the prediction of departure flights delays is realized.The main work is as follows:Firstly,proceeding from the flight operation process and flight status transition,a large-scale flight delay propagation model is established using the infectious disease model for reference.Through example analysis,the development and changing process of delay propagation and the affection of each transition parameter is analyzed.Then,the delay propagation within and between the airports is studied.Considering the actual operation process of flights,the intra-airport and inter-airport delay propagation routes and the corresponding dynamic propagation parameters are proposed.The intra-airport and inter-airport delay propagation models are established based on the state space theory.Through example analysis,the differences and different propagation routes of delay propagation in and between airports are analyzed.Finally,the real-time prediction of departure flights delays is studied.Considering the impact of weather,an airport weather impact index based on WITI is introduced.The BP neural network is used to predict the departure flight delays.70% flight data is selected for training,and the remaining 30% data is selected for prediction.The prediction results show the reliability of BP neural network.