飞行保障架次预测方法研究及恢复期预测  

Research on Prediction Method of Flight Guarantee Sorties and Prediction of Recovery Period

作　　者：茅佳宁 丁松滨[1] 刘计民 宋晓敏 MAO Jianing;DING Songbin;LIU Jimin;SONG Xiaomin(College of Civil Aviation,Nanjing University of Aeronautics and Astronautics,Nanjing 210016;CAAC East China Regional Administration,Shanghai 200335)

机构地区：[1]南京航空航天大学民航学院,南京210016 [2]中国民用航空华东地区空中交通管理局,上海200335

出　　处：《计算机与数字工程》2024年第5期1563-1568,共6页Computer & Digital Engineering

摘　　要：为研究飞行保障架次未来恢复发展情况,在传统时间序列预测方法基础上引入支持向量机(SVM)进行优化,再结合疫情影响预测并判断未来的增长情况,为未来航空运输的恢复提供了一定参考依据。首先基于ARIMA-SVM、Holt-Winters三参数指数平滑-SVM两种组合模型,在无疫情数据基础上进行验证,实现模型精度的优化;然后基于X-12分解疫情时间序列,预测2021年-2023年三年内的月度值,并判断年度增长恢复情况。结果表明:引入SVM优化残差序列后,组合模型与单一模型相比误差有所降低;通过疫情影响分析及预测可以判断疫情影响下的飞行保障架次预计在2023年恢复至疫情前的水平。In order to study the future recovery and development of flight guarantee sorties,this paper introduces support vec-tor machines(SVM)based on traditional time series forecasting methods to optimize,and then predicts and judges future growth based on the impact of the COVID-19 epidemic,which provides some reference for the recovery of air transportation in the future.Firstly,based on two combined models of ARIMA-SVM and Holt-Winters three-parameter exponential smoothing-SVM,the accu-racy of the model is optimized based on no epidemic data.Then by using X-12 to decompose epidemic time series,this study pre-dicts the monthly value within 2021-2023,and judges the annual growth recovery.The results show that using SVM to optimize the residual sequence,the error is reduced compared with the single model.Through the analysis of the epidemic impact,it can be judged that the flight guarantee sorties under the influence of the epidemic is expected to return to the level before the epidemic in 2023.

关 键 词：飞行保障架次 时间序列预测 ARIMA乘法季节模型 Holt-Winters三参数指数平滑 恢复期预测 

分 类 号：U8[交通运输工程]