基于GM-Markov模型的南宁吴圩国际机场货邮吞吐量预测

Forecasting cargo and mail throughput of Nanning Wuxu International Airport based on GM-Markov model

为帮助广西南宁吴圩国际机场科学制定货运相关政策并为机场管理部门提供科学的数据支持,以南宁吴圩国际机场2011—2022年货邮吞吐量数据为基础,构建序列累加生成数列,建立预测模型并进行模型误差分析及精确度检验。研究将灰色理论模型的相对误差划分为5个状态区间,分别应用灰色预测模型和灰色马尔科夫(GM-Markov)模型预测南宁吴圩国际机场2011—2022年的货邮吞吐量,并与实际吞吐量进行比较。研究表明:使用灰色理论模型,南宁市吴圩国际机场2011—2022年的货邮吞吐量预测平均误差为0.04;应用GM-Markov的方法,预测平均误差为0.008。显然GM-Markov预测模型要比传统的灰色预测模型的精度高,能大幅降低波动性较大的时间序列的预测误差,尤其适用于中短期的预测。由此根据GM-Markov预测模型,计算得出南宁吴圩国际机场2023—2026年的货邮吞吐量预测值。

The purpose of this essay is to ensure the scientific formulation of cargo-related policies and provide scientific data support for airport management at Nanning Wuxu International Airport in Guangxi.Based on the cargo and mail throughput data of Nanning Wuxu International Airport from 2011 to 2022,a serial accumulation series was constructed to generate a prediction model,a model error analysis and accuracy test were conducted.The relative errors of the gray theory model were divided into five state intervals,and the gray prediction model and the gray Markov model were applied to predict the cargo and mail throughput of Nanning Wuxu International Airport from 2011 to 2022.The grey prediction model and grey Markov model are applied to predict the cargo and mail throughput of Nanning Wuxu International Airport from 2011 to 2022,and are compared with the actual throughput.The study shows that the average error in forecasting cargo and mail throughput of Nanning Wuxu International Airport from 2011-2022 is 4%using the gray theoretical model.Applying the gray Markov approach,the average error of the forecast is 0.8%.Obviously,the grey Markov forecasting model has higher forecasting accuracy than the traditional grey forecasting GM model,which can significantly reduce the forecasting error of volatile time series,especially for short-and medium-term forecasting.Therefore,based on the gray Markov forecasting model,the forecasted cargo and mail throughput of Wuxue International Airport for 2023—2026 is calculated.