基于数字孪生的航班延误时间预测方法

Flight Delay Time Prediction Method Based on Digital Twins

为了更加准确和高效地预测大面积航班延误时间,提出了基于数字孪生的航班延误时间预测方法。首先,从航班链整体的角度出发,依据航班运行业务特点和数字孪生技术特征设计了航班链数字孪生系统框架,综合航班链全生命周期内相关航班和机场的运行状态特征;其次,基于Fastformer和GraphSAGE模型设计了航班链时空特征提取模型(ST-Former),充分挖掘航班之间的时空关联特征。实验表明,该方法预测效率和准确度显著提升,平均预测误差在3 min左右。

In order to predict flight delays in large areas more accurately and efficiently,a flight delay prediction method based on digital twins is proposed.First,from the perspective of the flight chain as a whole,a flight chain digital twin system framework was designed based on the flight operation business characteristics and digital twin technology characteristics,integrating the operating status characteristics of relevant flights and airports during the entire life cycle of the flight chain.Secondly,a flight chain digital twin system framework was designed based on the Fastformer and GraphSAGE models designed a flight chain spatio-temporal feature extraction model(ST-Former)to fully explore the spatio-temporal correlation features between flights.Experiments showed that the prediction efficiency and accuracy of this method were significantly improved,and the average prediction error was within 3 minutes.