Telemetry fault detection for meteorological satellite based on PCA

Satellite telemetry of meteorological satellite plays an important role for satellite safety management. In existing system, threshold value is used to monitor the abnormity because of large volumes of data which is collected every three seconds. However, the abnormal data in the threshold range cannot be detected timely. Based on this, in the paper we introduce Principle Component Analysis (PCA) method to detect the abnormity of satellite telemetry. Meanwhile, monitoring charts based on Hotelling's (T 2 ) and Square Prediction Error (SPE) statistics are used to decide whether a fault happens or not. Experiments show that PCA reports a state-of-the-art performance on satellite telemetry data.