Space event detection method based on cluster analysis of satellite historical orbital data

Abstract

A new method of detecting space events using satellite two-line element (TLE) histories is proposed. Anomalous data segments in the TLE-derived time series of a specific orbital element is detected to locate space events. After data preprocessing, a series of equal-length data segments are extracted from the time series and converted into a data form that is uniformly sampled in the time domain. Anomaly detection is achieved by clustering the data segments using a one-dimensional self-organizing map. After the clustering operation, the same type of data segments are grouped together and different types of data segments are divided into different groups. Thus, the type of space event and the associated orbital anomaly pattern can be obtained simultaneously. Space event detection results of typical active satellites show that the proposed method can accurately avoid false detections while maintaining a high detection rate. By comparing detection results of different clustering granularities, the ability to obtain detailed information of space events is also proved.