The Research On The Application Of Spacecraft Anomaly Detection Technology Based On Telemetry Series Data

In-orbit spacecraft is affected by the working environment and their own state,often happen various subsystems or components abnormalities.telemetry data from each sensor to the state of the acquisition on star,can reflect the working state of the spacecraft,so the state of satellite remote sensing temporal data processing and analysis is to understand,the only and effective means to administer a spacecraft in orbit.Aiming at the existing threshold range based on telemetry parameters,or the combination of multi-parameter threshold knowledge to detect the anomalies of spacecraft,there are some problems,such as the established threshold knowledge can’t completely cover all kinds of abnormal faults,and the abnormal changes of satellites within the threshold can’t be found effectively.In this paper,the anomaly detection method within the threshold based on telemetry timing data is studied.It is of great significance to detect the small changes or anomalies of spacecraft within the threshold in time for mastering the law of abnormal evolution and taking timely measures to prevent the occurrence of major faults.In this paper,the regular changes of parameter morphology of telemetry data during satellite operation are studied.It is found that some parameter changes show orbital periodicity,and have obvious morphological characteristics in the period,corresponding to the working state of the satellite,but usually the satellite will have a limited type model.In this paper,the telemetry timing data is taken as the research object.Firstly,based on the idea of telemetry timing data management and fast extraction to lay the data foundation for telemetry timing data analysis,the distributed storage concept of big data is fully applied,the storage structure of telemetry parameters is reconstructed,and the fast sequential access of telemetry data is realized.On the basis of the fast data access,based on the above-mentioned conclusion,a morphological change characteristic representation mode based on the grid division is proposed,and the calculation and extraction of the change-form characteristic of the telemetric data are realized;the segmentation of the sub-window of the data is completed according to the satellite orbit period,And the characteristics of the telemetering historical data are realized.The method of clustering realizes the establishment of the parameter mode template,realizes the detection of the quasi-real-time parameter abnormal period based on the template matching,and realizes the detection of the abnormal period existing in a section of historical data based on the space search algorithm,Finally,by calculating the correlation coefficient matrix of the multi-parameter in the abnormal period,the cluster of the parameters in the abnormal period is found,and the abnormal knowledge of the existing spacecraft is used,and the corresponding relation between the data abnormal period and the spacecraft’s abnormality is preliminarily realized.Through the test of actual telemetry data,it is found that the automatic abnormal period detection of nearly 20% of the characteristic parameters of orbit period can be completed automatically by using this method.Through parameter clustering and using the existing fault mode library,the abnormal period is mapped to the anomalies of satellite subsystems or devices,and the correlation mapping between data anomalies and spacecraft anomalies is preliminarily realized,which is of great significance to spacecraft anomaly detection.It promotes the application of telemetry data in spacecraft management.