A Data-driven System-level Health State Prognostics Method for Large-scale Spacecraft Systems

Abstract

Large-scale spacecraft, such as space station, highlights the systems’ reliability and safety. Using prognostics to predict the trend of the system health state evolution can help find out the potential dangers and prevent the unexpected failure from happening. With the adoption of data-driven ideology, a system-level health state prognostics method is proposed to predict the trend information. First, the characteristics of the large-scale spacecraft and the system-level health definition are analyzed. Then the details of the solution method are described. The novelty of this method is to use the network science knowledge to extract the system-level features. The adopted predicting method is briefly introduced. Finally, a real case study with on-orbit telemetry data is presented, and relevant conclusions are drawn for reference.