基于小波神经网络的飞行到达时刻预测研究  

Research on Flight Arrival Time Prediction Based on Wavelet Neural Network

作　　者：郑方圆 张光明[1] Zheng Fangyuan;Zhang Guangming(Civil Aviation Flight University of China,Guanghan 618307)

机构地区：[1]中国民用航空飞行学院,广汉618307

出　　处：《现代计算机》2022年第3期13-18,共6页Modern Computer

摘　　要：航空器的预测到达时刻(estimatedtimeofarrival,ETA)是间隔管理的基础。针对提高航空器的到达时刻的预测准确性,提高空中飞行容量的同时减小冲突,提出了基于ADS-B数据的小波神经网络预测方法。先将本机的原始ADS-B数据进行预处理,使用小波神经网络方法构建预测模型进行预测;同时获取前机空中广播的最新的ADS-B数据进行增量学习,实时辅助获取前方突发机动飞行空域的最新状况,提高模型的预测能力。最后以某航班为例建立仿真实验,结果显示预测到达时刻的平均绝对误差在15 s内,表明无论是在稳定状况还是突发状况,小波神经网络都有较高的准确性和稳定性。Estimated time of arrival(ETA)of aircraft is the basis of interval management.Aiming at improving the prediction accuracy of aircraft arrival time,improving air flight capacity and reducing conflict,a wavelet neural network prediction method based on ADS-B data is proposed.Firstly,the original ADS-B data of the aircraft is preprocessed,and the wavelet neural network method is used to build a prediction model for prediction Obtain the latest ADS-B data from the front aircraft air broadcast for in⁃cremental learning,assist in obtaining the latest situation of the front sudden maneuver flight airspace in real time,and improve the prediction ability of the model.Finally,taking a flight as an example,the simulation experiment is established.The results show that the average absolute error of the ETA is within 15 s,which shows that wavelet neural network has high accuracy and stability in both stable and sudden conditions.

关 键 词：预测到达时间 小波神经网络 ADS-B 数据挖掘 

分 类 号：V355[航空宇航科学与技术—人机与环境工程] TP183[自动化与计算机技术—控制理论与控制工程]