基于气象因素的机场进离港延误预测  

Airport arrival and departure delay time prediction based on meteorological factors

作　　者：姜雨[1] 袁琪 胡志韬 吴薇薇[1] 顾欣 JIANG Yu;YUAN Qi;HU Zhitao;WU Weiwei;GU Xin(College of Civil Aviation,Nanjing University of Aeronautics and Astronautics,Nanjing 211106,China;Beijing Key Laboratory of Traffic Engineering,Beijing University of Technology,Beijing 100124,China)

机构地区：[1]南京航空航天大学民航学院,江苏南京211106 [2]北京工业大学北京市交通工程重点实验室,北京100124

出　　处：《系统工程与电子技术》2023年第6期1722-1731,共10页Systems Engineering and Electronics

基　　金：国家自然科学基金(U1933118,U2033205)资助课题。

摘　　要：针对机场延误预测过程中难以提取延误传播时空特征、预测结果受天气扰动大的问题,提出了基于气象因素的时空图卷积网络(meteorology-based spatio-temporal graph convolutional networks, MSTGCN)机场延误预测模型。该模型使用图卷积神经网络(graph convolutional neural network, GCNN)与门控卷积神经网络(gated convolutional neural network, Gated CNN)挖掘机场延误的时空特征,同时加入气象特征提取模块对机场延误时间进行预测。实验结果表明,该模型在中短时预测上的表现均优于其他对比模型;相较于不考虑气象因素的模型,MSTGCN对未来1 h、4 h和12 h预测的平均绝对误差分别降低了7.03%,7.93%,11.54%,对预测结果起到了极大的修正作用。In order to solve the problem of great effect by weather conditions and the difficulty in extracting the temporal and spatial features of delay in the process of airport delay prediction,a meteorology-based spatio-temporal graph convolutional networks(MSTGCN)model is proposed in this paper.This model uses graph convolutional neural network(GCNN)and gated convolutional neural network(Gated CNN)to extract the temporal and spatial features of airport delays,and integrates meteorological characteristics extracting module to predict delay time of airports.The experiment results show that the performance of the proposed model in medium and short term prediction is superior to other comparative models.Compared with the model that does not consider meteorological factors,the mean absolute errors of MSTGCN for the next 1 h,4 h and 12 h are respectively decreased by 7.03%,7.93%,11.54%,which greatly revises the prediction results.

关 键 词：机场延误预测 图卷积神经网络 气象因素 机场网络 深度学习 

分 类 号：V351[航空宇航科学与技术—人机与环境工程]