基于K-means++时间聚类的飞行航迹预测  

Flight Path Prediction Based on K-means++Time Clustering

作　　者：郭宪超 李廷元[1] Guo Xianchao;Li Tingyuan(College of Computer Science,Civil Aviation Flight University of China,Guanghan 618307)

机构地区：[1]中国民用航空飞行学院计算机学院,广汉618307

出　　处：《现代计算机》2022年第3期54-57,共4页Modern Computer

摘　　要：针对传统航迹聚类预测算法对时间数据预测不准确的问题,本文提出了一种基于K-means++时间聚类的航迹预测算法。该算法使用ADS-B数据作为数据源,通过K-means++算法对时间数据进行聚类,使用修正欧式距离算法进行航迹相似度的计算,分别根据直线阶段和转弯阶段的特征分配不同的聚类方法得出聚类航迹,最后对聚类航迹进行当前位置修正得到预测航迹。通过实际航迹数据进行算例试验,结果显示该预测算法具有较高的准确性。Aiming at the problem that the traditional track clustering prediction algorithm can not predict the time data accu⁃rately,a track prediction algorithm based on K-means++time clustering is proposed.The algorithm using ADS-B data as data sources,through the K-means++clustering algorithm for time data,using the modified Euclidean distance similarity computing al⁃gorithm to track,respectively,according to the features of straight line and turn phase distribution of the different clustering method of track,finally,the clustering of track is to obtain the forecast track is the current position.An example test is carried out with the actual flight path data,and the results show that the prediction algorithm has high accuracy.

关 键 词：聚类算法 K-means++ 航迹预测 

分 类 号：V355.1[航空宇航科学与技术—人机与环境工程] TP311.13[自动化与计算机技术—计算机软件与理论]