遥感卫星在轨故障挖掘分析

Data Mining and Analysis in Satellite On-orbit Faults of Meteorological and Remote Sensing Satellites

为评估传统航天领域卫星总体的最新设计特点及研制水平,提出基于时空分布特征及相关性的故障数据挖掘分析方法。针对2004年至2021年期间某卫星总体抓总设计研制的气象领域、通信试验领域等多领域遥感卫星的在轨故障数据,依次研究了分系统随时间变化引发的故障,分系统故障类型的多维变量故障率及跨系统之间的故障相关性。结果表明,控制分系统故障率最高,其中空间环境影响最大,并与测控分系统故障变化趋势存在强相关性。现有的技术性分析方法集中于在轨卫星故障的诊断预测,而基于时空分布特征的故障相关性挖掘分析方法聚焦组织内部卫星各分系统的故障特点,有利于吸取可靠性设计经验和教训,积累组织过程资产,有助于提高卫星总体设计水平和卫星项目质量管理水平。

In order to evaluate the latest design features and development level of satellite in the traditional aerospace field, a fault correlation mining analysis method based on time-space distribution features is proposed. Aiming at the on orbit fault data of remote sensing satellites in meteorological field,communication test field and other areas developed by satellite overall master design from 2004 to 2021, the faults caused by subsystem changes with time, multi-dimensional variable fault rate of subsystem fault types and fault correlation between cross systems are studied in turn. The results show that the control subsystem has the highest failure rate, in which the space environment has the greatest impact, and there is a strong correlation with the fault change trend of the measurement and control subsystem. The existing technical analysis methods focus on the diagnosis and prediction of on orbit satellite faults, while the fault correlation mining analysis method based on the time-space distribution features focuses on the fault characteristics of the internal satellite subsystems, which is conducive to learning the experience and lessons of reliability design, accumulating organizational process assets, and improving the overall satellite design level and satellite project quality management level.