Research On Assessment Methods Of Health Status For On-orbit Satellite

As the number of China’s satellites is increasing year by year,on-orbit satellite monitoring system has difficulty in adapting to the current requirement.Therefore,the assessment methods of health status for on-orbit satellites are of great significance to the health management and high reliability of spacecraft during on-orbit operation.In order to solve this problem,which involves depicting the complex redundant,the reconfigurable and nonlinear relationships among the satellite systems and utilizing the large number of satellite telemetry data,it is necessary to study the assessment methods of health status for multi-level on-orbit satellite from the data-driven perspective.It mainly includes the following aspects:Firstly,methods of feature extraction that can indicate the health status of satellite components are studied accroding to the characteristics of satellite telemetry data,which is of discrete value,lager quantity of data,complex anomalies,etc.Then,the fluctuation features are extracted during this process.As an application,the validity of these features are verified under the method of anomaly detection based on sequential probability ration test,fluctuation features and the actual telemetry data.Secondly,this study puts forward assessment methods of health status of satellite components based on observational residuals which are obatained through the comparison of actual data and historical data and regarded as health status features.Then this paper raised two methods based on sequential probability ration and non parametric regression method respectively to conduct the fusion of health status features of components.As a result,their health index value are gained during the process.Thirdly,this paper puts forward two novel kinds of assessment methods of health status for on-orbit satellite systems.The first method is based on reconfigurable degree,which can portray the complex,reconfigurable and nonlinear characteristics of the satellite systems.Analogously,the second method is based on the Member Function Map of satellite systems,which describes the characteristics of satellite systems.Fourthly,through the actual telemetry data,both of above two methods are verified,and the results show that the proposed two novel methods are of high performance in algorithm efficiency,real-time,interpretability and practicality.Finally,the research contents are summarized and the future research contents are prospected.