Application of sparse representation method based on K-SVD-ADMM in anomaly detection of satellite telemetry

In aerospace, anomaly detection based on telemetry data is a major task satellite health monitoring that is important for identifying anomalous events and for taking measurements to improve system reliability and safety. In this paper, a novel anomaly detection model using sparse representation method based on K-means Singular Value Decomposition (K-SVD) and Alternating Direction Method of Multipliers (ADMM) is introduced. The proposed K-SVD-ADMM is used to solve optimization problems and reconstruct the telemetry data. The anomaly score between the reconstructed and observed values is calculated and an adaptive threshold method is proposed to detect anomalies. Case analysis of satellite antenna component telemetry data shows that the proposed method is more stable and accurate compared with other existing anomaly detection methods based on One-class support vector machine (OCSVM) and Long Short-Term Memory (LSTM).