一种开放式通用小卫星自主健康检测系统  

An open general autonomous health monitoring system for small satellites

作　　者：苏晨光 徐婧瑶 刘一鹤 李源 李军予[1] SU Chenguang;XU Jingyao;LIU Yihe;LI Yuan;LI Junyu(DFH Satellite Co.,Ltd.,Beijing100094,China)

机构地区：[1]航天东方红卫星有限公司,北京100094

出　　处：《先进小卫星技术(中英文)》2025年第1期70-77,共8页Advanced Small Satellite Technology

摘　　要：针对采用高性能处理器及高级通用操作系统的小卫星,提出了一种开放式通用小卫星自主健康检测系统架构,利用数据分发服务(data distribution service,DDS)技术将卫星数据与检测模型解耦,能够支持多种独立检测算法的并行运算与“即插即用”。在检测模型中引入了包括深度学习算法在内的多种无监督机器学习算法,通过选取一定时间内卫星状态相对健康的遥测数据进行离线训练,再将训练好的模型加载至系统,从而实现对卫星健康状态的实时检测.利用某在轨型号的历史遥测数据进行模拟验证,验证结果表明:1)该检测系统具备良好的扩展性与伸缩性;2)无监督机器学习算法具备在卫星自主健康检测领域的应用条件.An open autonomous health monitoring system architecture was presented for small satellites equipped with high performance processors and advanced general operating systems.Data distribution service(DDS)technology was used to decouple the satellite data from the monitoring models,allowing parallel execution of different monitoring algorithms and supporting plug-and-play functionality.The system was integrated with a set of unsupervised machine learning algorithms,including the deep learning one,and trained offline using the telemetry data for a period of time from the healthy satellite state.This trained model was then used in the realtime satellite health monitoring system.Simulation validation with historical telemetry data of an on-orbit satellite model confirms the system's scalability,flexibility,and the applicability of unsupervised machine learning algorithms in autonomous satellite health monitoring.

关 键 词：卫星自主健康检测 数据分发服务 机器学习 无监督学习 数据驱动 

分 类 号：V474[航空宇航科学与技术—飞行器设计]