基于机场子区划分的机场网络交通拥挤传播预测

为了预测机场进离场交通的拥挤态势,本文从机场网络的角度进行研究.首先针对交通拥挤形成的动态过程,建立了基于出入流率的交通拥挤的定义及其度量;接着,引入多维标度法对机场之间的交通相关性进行定量分析,划分机场子区,以降低网络分析的复杂度及解空间维数;然后,构建了基于Elman神经网络实现机场子区内多个相关机场的交通拥挤传播预测方法;最后,基于美国机场的实际航班数据对机场网络拥挤传播预测方法进行验证.验证结果表明,预测结果的平均绝对百分比误差和平均绝对偏差较小,明显优于对比算法.

Forecasting of Airport Network Traffic Congestion Propagation Based on Division of Airport Subareas

In order to predict congestion of airport arrival and departure traffic, the airport network is studied. Definition and measurement of airport congestion is established firstly based on inflow and outflow rates, which describes the formation process of traffic congestion. Secondly, Multidimensional scaling theory is applied to study the relationship of airports and divide the whole airport network into airport subareas, in order to simplify the complexity of airport network analysis, and to reduce the dimension of solution space. Thirdly, the method of congestion propagation forecasting of airport subareas based on Elman neural network is introduced. In the last section, verification was performed using real flight data of US airports. The results indicate that the Mean Absolute Percentage Error and Mean Absolute Deviation small. The method is proved to be super to BP neural network.