关键词: 航空运输, 客运量预测, 三元区间数, 区间时间序列, 灰色周期外延模型

Abstract: To solve the forecasting problem of the stochastic passenger flow volume with the characteristic of poor information, small sample and non-informative prior distribution pattern, the upper and lower bound information of the passenger flow time series of these routes were extracted and a preference value was added into the middle to form a ternary interval number structure. This structure contained three elements, namely, left boundary point, middle point and right boundary point. Then this ternary interval number data structure was transformed into three independent time series, namely, a left and a right radius time series as well as a center time series. The gray system theory was used to establish a route passenger flow prediction model, and the gray extension model was then used to compensate the residual sequence obtained by the above model. The passenger transportation volume data of civil aviation ranging from 2004 to 2019 was used to conduct the verification analysis. The result shows that the mean absolute percentage error (MAPE) resulting from ARIMA method is 6.77% as opposed to 1.66% from the use of the grey-periodic extensional combinatorial model. Therefore, the latter has a significant advantage in short-term prediction.

Key words: air transportation, passenger traffic volume forecast, ternary interval number, interval time series, grey-periodic extensional combinatorial model