摘要
造成航班延误有很多潜在且不确定的因素,目前尚未有方法可以有效地避免航班的大面积延误。论文基于GLO机场延误特性的分析为基础,筛选与机场航班离港延误时长相关的因素并添加周期性变量共计13个特征变量建立训练集进行深度学习。利用RBFNN、BPNN、WNN,进行对比仿真实验,仿真结果表明考虑周期性波动因素的模型较原模型预测准确度提高,其中RBF神经网络模型预测准确度最高,仿真结果:±10min、±5min、±3min容差内延误时长预测准确率分别为98%、94%、91%。
There are many potential and uncertain factors that cause flight delays.At present,there is no way to effectively avoid large-scale flight delays.Based on the analysis of GLO airport delay characteristics,this article selects factors related to the length of airport flight departure delays and adds cycles.A total of 13 characteristic variables are used to establish a training set for deep learning.By using RBFNN,BPNN and WNN,comparative simulation experiments are carried out,and the simulation results show that the model considering periodic fluctuation factors have higher prediction accuracy than the original model.Among them,the RBF neural network model has the highest prediction accuracy.The simulation results show that the prediction accuracy of delay time within±10min,±5min,±3min tolerance is 98%,94%and 91%respectively.
作者
张启凡
王永忠
王圣堂
裴柯欣
ZHANG Qifan;WANG Yongzhong;WANG Shengtang;PEI Kexin(School of Air Traffic Control Management,Civil Aviation Flight University of China,Guanghan 618300)
出处
《舰船电子工程》
2021年第7期133-136,共4页
Ship Electronic Engineering
基金
国家级大学生创新创业训练计划项目(编号:201910624041)资助。
