跑道占用时间的影响因素分析与预测方法研究

Research on influencing factors analysis and predicting method of runway occupancy time

针对最后进近阶段飞机距跑道入口较远时,跑道占用时间难以实现准确且实时的预测,提出了基于机器学习的跑道占用时间预测方法。面向空中交通管理,考虑3类共10项影响因素与跑道占用时间的关系;然后,基于对影响因素的分析,使用BP神经网络建立按照飞机尾流等级分类的跑道占用时间预测模型;通过对模型的调试和训练,对模型性能、影响因素显著性和不同距离时的预测结果展开了分析。研究表明,模型对跑道占用时间的估算平均绝对误差不超过2.5 s,均方根误差不超过3 s,判定系数均在0.85以上,可以很好地满足跑道占用时间的准确预测;影响因素偏差导致的估算误差随着飞机尾流等级的减小而增大;当飞机位于距跑道入口19 km时,估算误差为4.33 s,是预测跑道占用时间的较佳位置。

Considering the difficulty to predict runway occupancy time accurately and in real time when the aircraft is far away from the runway threshold in final approach,this paper proposed a run-way occupancy time prediction method based on machine learning.From the perspective of air traffic management,the relationship between 10 influencing factors in 3 categories and runway occupancy time was considered.Then,based on the analysis of the influencing factors,the BP neural network was used to establish the runway occupancy time prediction model according to the aircraft wake lev-el.After debugging and training of the model,the performance of the model,the significance of in-fluencing factors and the prediction results at different distances were analyzed.The results show that the average absolute error of runway occupancy time estimation by the model is less than 2.5 s,the root-mean-square error is less than 3 s.and the coefficients of determination are all above 0.85,which can well meet the accurate prediction of runway occupancy time.The estimation error caused by the deviation of influencing factors increases with the decrease of aircraft wake level.When the aircraft is located at 19 km away from the runway entrance,the estimation error is 4.33 s,which is a better position to predict the runway occupancy time.