Satellite System Anomaly Detection Method Based on Intelligent Data Mining Technology

With the increasing complexity of advanced satellite functionality and composition, there is a greater variety of abnormal phenomena that can occur during satellite operations. To avoid a significant increase in modeling work for satellite anomaly detection, an intelligent-based method for satellite system anomaly detection is proposed. This method utilizes transfer learning to achieve efficient analysis of correlated parameters. A subsystem-based multi-parameter anomaly detection scheme based on semi-supervised learning is proposed to address challenges such as a lack of satellite data labels and non-linearity of data. Deep learning techniques are employed to fully utilize the characteristics of time series data and accomplish comprehensive system-wide anomaly detection.