基于改进KNN检测的ADS-B轨迹插补研究  

Research on ADS-B trajectory interpolation based on improved KNN detection

作　　者：靳慧斌 冯朝辉 张召悦[2] 王志森 IN Huibin;FENG Chaohui;ZHANG Zhaoyue;WANG Zhisen(College of Safety Science and Engineering,CAUC,Tianjin 300300,China;College of Air Traffic Management,CAUC,Tianjin 300300,China)

机构地区：[1]中国民航大学安全科学与工程学院,天津300300 [2]中国民航大学空中交通管理学院,天津300300

出　　处：《中国民航大学学报》2023年第4期23-28,共6页Journal of Civil Aviation University of China

基　　金：中国民航大学研究生科研创新基金项目(10502756)。

摘　　要：针对广播式自动相关监视(ADS-B,automatic dependent surveillance broadcast)地面接收站数据存在缺失点、离群点等异常点的问题,在建立多约束条件判别模型的基础上筛选ADS-B轨迹缺失点,并对航迹进行插值,采用标准化欧氏度量的改进邻近算法对整条轨迹进行异常点数据检测,最后应用多重插补法对异常轨迹点进行修复。以实际运行数据为例,验证对比了改进的K最近邻(KNN,K-nearest neighbor)算法用于检测ADS-B异常轨迹点的准确率,该方法可有效解决轨迹异常点检测疏漏问题,为航空器ADS-B原始轨迹数据挖掘提供支撑。Aiming at the problem of anomalous points such as missing points and outliers in the data of automatic dependent surveillance broadcast(ADS-B)ground receiving station,the missing points of ADS-B trajectory are screened on the basis of establishing a multi-constraint discriminant model,and the trajectory is interpolated.The improved neighbor algorithm with standardized Euclidean metric is used to detect anomalous point data of the whole trajectory.Finally,the multiple interpolation method is applied to repair the anomalous trajectory points.Taking actual operational data as an example,the accuracy of the improved K-nearest neighbor(KNN)for detecting ADS-B anomalous trajectory points is verified and compared.This method can effectively solve the problem of detection omission of trajectory anomalous point and provide support for mining original trajectory data of aircraft ADS-B.

关 键 词：ADS-B数据筛选 异常点检测 标准化欧氏度量 改进邻近算法 多重插补方法 

分 类 号：V355[航空宇航科学与技术—人机与环境工程]