Research On Forecasting Of Airline Passenger Flow Volume In Small And Medium-sized Airports

Carrying out airline passenger flow forecasting is the premise for small and medium-sized airports to plan and design,develop and operate airline,and measure market operation subsidies.Therefore,this thesis establishes the topic of "Research on Airline Passenger Flow Forecasting of Small and Medium-sized Airports" to provide theoretical support for small and medium-sized airports to make relevant market decisions through the research on this topic.According to the characteristics of airline operations in small and medium-sized airports,the airline passenger flow forecasting for small and medium-sized airports is divided into two types: existing airport airline and newly opened airline.Therefore,the main research work carries out in this thesis is as follows:(1)The problem of passenger flow forecasting of existing routes that lack production and operation data in small and medium-sized airports is studied.For the time series composed of route passenger flow volume with the characteristics of overall changing trend and periodic fluctuation,this thesis constructs a GM(1,1)model to forecast the trend of the series.Through obtaining the residual sequence of the predicted value of the gray model,a periodic extension model is presented to extract the dominant period of the residual sequence,and then a combination of the gray periodic extension model is constructed to predict the periodic fluctuation.After that,the residual error of gray periodic extension model is corrected and compensated based on the support vector machine model.Empirical analysis finds that this model has high accuracy.(2)The problem of passenger flow forecasting of newly opened routes in small and medium airports is studied.For the problem of passenger flow forecasting of new routes without production and operation data,this thesis presents an index system of related factors impacting the size of passenger flow volume.Both calculating approaches of spatial distance and direct similarity are proposed,such that being applied to calculate the similarity value between target and reference sample in which 2-D attribute indices are involved.Then the coefficient of variation is used to allocate weights to each index so as to obtain the reference sample set of objective samples.Stepwise regression analysis is used to select explanatory variables that have a significant impact on explained variables.After that,the selected explanatory variables are used as input neurons and the passenger flow as output neurons.Then BP-neural network is used to train the reference sample set and obtain the passenger flow of new routes forecast result.The case analysis shows that the model has high forecast accuracy and can provide decision support for passenger flow forecast of new routes.