基于时间序列的多维距离聚类异常检测方法  

Multi-dimensional distance clustering anomaly detection method based on time series

作　　者：丁建立[1] 黄天镜 徐俊洁 王静[1] DING Jian-li;HUANG Tian-jing;XU Jun-jie;WANG Jing(College of Computer Science and Technology,Civil Aviation University of China,Tianjin 300300,China)

机构地区：[1]中国民航大学计算机科学与技术学院,天津300300

出　　处：《计算机工程与设计》2020年第7期1935-1940,共6页Computer Engineering and Design

基　　金：民航安全能力建设基金项目(AADSA0018);中国民航大学人事处基金项目(107/10701004)。

摘　　要：目前轨迹分析的异常行为检测技术以检测位置信息为主,忽略了时空轨迹的轨迹有序性及运动特性,为此提出一种基于时间序列的多维特征聚类异常检测方法,提高广播式自动相关监视(ADS-B)数据异常检测技术的精确性,通过提取ADS-B数据中经度、纬度、速度、航向信息,利用Hausdorff距离计算轨迹数据的多特征相似度,结合层次聚类方法检测轨迹中的异常行为。实验结果表明,该方法能够有效提高飞行轨迹数据的异常行为检测的精确性。The abnormal behavior detection technology for the current trajectory analysis mainly detects the position information,ignoring the trajectory order and motion characteristics.A time-based multi-dimensional feature anomaly clustering detection method was proposed to improve the accuracy of the automatic dependent surveillance-broadcast(ADS-B)data anomaly detection technology.By extracting the longitude,latitude,speed and heading information of ADS-B data,the Hausdorff distance was used to calculate the multi-feature similarity of the trajectory data,and the hierarchical clustering method was combined to detect the anomalous behavior in the trajectory.Experimental results show that the proposed method can effectively improve the accuracy of abnormal behavior detection of flight trajectory data.

关 键 词：广播式自动相关监视 多维特征 HAUSDORFF距离 层次聚类 异常检测 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]