基于数据挖掘的在轨卫星故障预测分析  

Fault Prediction and Analysis of On-orbit Satellites Based on Data Mining

作　　者：钱昭勇 曹裕华 史增凯 张雷[3] QIAN Zhaoyong;CAO Yuhua;SHI Zengkai;ZHANG Lei(Space Engineering University,Beijing 101416,China;Joint Service College,National Defence University,Beijing 100858,China;Xi’an Satellite Control Center,Xi’an 710043,China)

机构地区：[1]航天工程大学,北京101416 [2]国防大学联合勤务学院,北京100858 [3]西安卫星测控中心,西安710043

出　　处：《火力与指挥控制》2022年第11期164-169,共6页Fire Control & Command Control

基　　金：军事类研究生资助课题(JY2019C213)。

摘　　要：对于在轨卫星的在役考核,其质量稳定性是一项非常关键的指标。一个重要的途径就是通过对在轨卫星的工作状态及健康状况保持持续监控和记录,充分利用历史各种类型的遥测告警数据,进行相关性检验,判断哪些类型的告警与对应卫星分系统的真正故障紧密相关。构建时间序列ARMA模型预测相关类型的告警情况,采用支持向量机(SVM)和随机森林(RF)等预测模型进行分类判断,进而预测在轨卫星是否发生某具体类型的故障,为检验考核卫星系统或部件质量稳定性提供参考方式。For the in-service assessment of on-orbit satellites,the quality stability is a very important index.An important way is continuously monitoring and recording the working state and health status of the on-orbit satellites,making full use of various types of historical telemetry alarm data,conducting the correlation test,and judging which type of alarm is closely related to the real fault of the corresponding satellite subsystem.The time series model ARMA is constructed to predict the relevant types of alarm conditions,then such prediction models as support vector machine(SVM)and random forest(RF),etc. are used for classification and judgment,and whether a specific type of fault occurs in the on-orbit satellites is predicted.It provides a reference method for the inspection and assessment of the quality stability of satellite system or components.

关 键 词：数据挖掘 在轨卫星 在役考核 时间序列 故障预测 

分 类 号：E92[军事—军事装备学]