Anomaly detection of satellite telemetry data based on extended dominant sets clustering

Abstract To mine out anomalies in satellite telemetry data under unsupervised conditions, a cluster-based method is proposed this paper. Firstly, an extended dominant sets clustering algorithm to cluster the with arbitrary shapes. Secondly, objects that do not belong any or small clusters are traditionally identified as anomalies. Thirdly, large detected according relative similarity. Finally, information on anomaly windows sequence obtained local rate, which provides more characteristics of Experimental results show that: 1) The can deal dataset containing multiple and arbitrarily shaped clusters; 2) introduction similarity increases AUC values detection by 3%~10%; 3) effectively detect magnetometer Tianping-2B satellite. Therefore, achieve provide support for improving safety reliability.