摘要
针对自旋稳定卫星遥测数据异常时,故障类型和故障源难于辨识的问题,提出一种数据驱动的故障诊断新思路。该方法提出将卫星姿态与传感器之间固有冗余关系是否保持作为判断故障类型的依据,并利用主成分分析把这种关系通过测量空间的特征值进行量化,监测特征值之间相应比例的变化实施判定;借助正交子空间的结构分析,选择SPE统计量进行故障检测和故障源的定位。在遥测数据的基础上模拟风云某型气象卫星的不同故障模式,并应用本文方法进行分析,结果表明,该方法克服了传统贡献图方法无法判别故障类型的缺点,可以实现对故障的正确诊断。
A new method for fault diagnosis is proposed to solve the identification problems of cause and source of fault when the attitude data of spin-stabilized satellite change abnormally.A new criterion for judging the cause of fault is proposed based on the inherent redundant relationship between satellite attitude and sensors,and this relationship is quantized by using eigenvalues of the measurement space by principal component analysis,then by monitoring the proportional change of eigenvalues,the cause of fault can be determined;after analyzing the structures of principal component subspace and residual subspace,the statistical variable SPE is chosen to detect fault and locate the fault source.Different fault modes are simulated on the basis of telemetriy data of the FY satellite.Compared with the traditional method,our results demonstrate that the proposed algorithm can identify the cause of fault and make a correct diagnosis.
出处
《宇航学报》
EI
CAS
CSCD
北大核心
2011年第6期1327-1332,共6页
Journal of Astronautics
基金
航天支撑基金(N9XW0002)
西北工业大学科技创新基金(2008KJ02011)
关键词
自旋稳定卫星
姿态传感器
数据异常
故障类型
故障源
Spin-stabilized satellite
Attitude sensor
Abnormal data
Fault type
Fault source
