文档介绍：南京航空航天大学
硕士学位论文
基于分类法的机场货运吞吐量预测方法研究
姓名:南娟
申请学位级别:硕士
专业:交通运输规划与管理
指导教师:夏洪山
20081201
南京航空航天大学硕士学位论文
摘要
长期以来,对机场货运吞吐量进行预测,一般是尝试多种预测方法后选取精度高的预测方
法或改进某一种预测方法,没有从理论上系统的阐述机场货运吞吐量的生成机理、影响因素以
及对于一个具体的机场,如何从上百种方法中选择适合的预测方法,特别是忽视了对于新建机
场和既有机场两种不同类型的机场,货运吞吐量预测的差异性,因而造成了机场货运吞吐量预
测的不准确性。
本文首先在分析了机场货运吞吐量的生成机理,影响因素,以及现有算法适用性的基础上,
给出了预测机场货运吞吐量选取的指标。然后对现有预测算法及其适用性进行了总结分析。针
对现有算法的不足,将熵权法引入了新建机场货运吞吐量的预测,给出了一种针对新建机场的
基于熵权法的综合评价预测方法,解决了新建机场没有历史数据很难进行定量预测这个问题。
通过九寨沟黄龙机场货邮吞吐量预测的实例分析,证明了该算法较以往算法具有较高的预测精
度;而对于现有的运营机场,给出了一种改进的组合预测方法。对于其中用到的计量经济法进
行了改进,运用因子分析法消除了变量之间的多重共线性,使新回归方程的回归系数通过了显
著性检验,并采用熵权法计算了组合预测中用到的各算法的权值。利用该算法对上海机场 2006
年货运吞吐量进行了预测,结果显示该预测算法也具有较高的预测精度。

关键词:民航货运,货运吞吐量预测,熵权法,因子分析,组合预测

I
机场货运吞吐量预测方法研究
Abstract
For a long time, air cargo throughputs are predicted by screening from a variety of methods
according to precision or improving certain method. There is little systematic exposition of their
generating mechanisms and influencing factors. What is more, how to select a proper method for a
given airport hasn't been solved in theory, especially for the differences between new airports and
existing ones, which leads to the forecasting uncertainty.
In this paper, firstly, the generating mechanism and influencing factors of air cargo throughputs
are analyzed. Based on the analysis, as well as existing algorithms and their adaptabilities, a set of
indexes is selected for throughput forecast. Then the existing algorithms and their applicability are
summarized and analyzed. To e the deficiency of the current algorithms, prehensive
evaluation index system based upon entropy method is introduced for new airports throughput
forecast, which solved the problem of quantitative predictions for new airports in lacking of historical
data. In a case study of freight throughput of Huanglong airport, this algorithm proves to be more
accurate. Besides, an bination forecasting algorithm is given to forecast the freight
vol