Anomaly detection of spacecraft based on least squares support vector machine

Anomaly detection is to identify abnormal behaviors of a system. It plays an important role in Prognostics and System Health Management (PHM) which is applied to monitor spacecraft´s performance, detect faults, identify the root cause of the fault, and predict the remaining useful life (RUL) in order to increase safety and reduce downtime of spacecraft in-orbit. This paper proposes a method to detect anomaly of spacecraft in-orbit based on Least Squares Support Vector Machine (LS SVM), which has excellent learning, classification and generalization ability. The process is as follows: (1) Data Collection and Preprocessing. Data such as current, voltage, temperature, vibration and so on are acquired in spacecraft in orbit and preprocessed firstly after they are sent to remote measurement and control centers in the ground, including filtering out the noise and removing non-cutting signals. (2) Feature Extraction. Features which contain any combination of numerical and categorical values are extracted and identified from the sensors´ signals using statistical methods. (3) Feature Selection. Principal Component Analysis (PCA) is used to select a subset of features that provides a more informative, robust representation of the information contained in the data since PCA is generally effective for dimensionality reduction and data compression. (4) Anomaly Detection. LS-SVM is used to identify abnormal behaviors of spacecraft in-orbit. Meanwhile, the time when the anomaly was first noticed and the potential causes of the anomaly ranging from subsystem to system level are determined. Finally, experimental data from a spacecraft in-orbit are used to test the performance of the algorithm and the results show that the proposed method achieves perfect accuracy and efficiency in anomaly detection of spacecraft in-orbit.