基于ILLE和SVM的卫星执行机构系统故障检测与定位

Fault Detection and Location of Actuator System in Satellite Based on ILLE and SVM

针对某超高指向精度要求的卫星平台,采用增量式局部线性嵌入(ILLE)与支持向量机(SVM)结合的方法,研究系统配置的多组磁伺服机构的故障检测与故障定位技术。在分析执行机构故障模式以及故障影响的基础上,采用LLE算法实时提取并更新与故障相关的卫星姿态控制系统高维信息,对其进行降维及特征提取,实现执行机构系统故障检测。当检测到故障时,提取执行机构系统输入输出信息,利用支持向量机(SVM)方法进行故障定位。该方法无需采集离线数据生成样本集,直接利用卫星姿控系统产生的在线故障特征数据集进行故障检测,并能根据故障检测结果,有效地实现卫星姿控系统执行机构的故障定位。算例仿真结果验证了所提方法的有效性。

In consideration of a satellite platform with requirements of ultra-high pointing accuracy,a fault diagnosis method based on incremental local linear embedding(ILLE)and support vector machine(SVM)is used to study the fault detection and fault location technology of multiple sets of satellite-borne magnetic servo mechanisms.Based on the analysis of the failure mode of the actuator and the impact of the failure,the LLE algorithm is used to extract and update the high-dimensional information of the satellite attitude control system related to the fault in real time,and the dimension and feature extraction is reduced to realize fault detection of the actuator system.When a fault is detected,a method based on SVM is proposed with input and output information of the actuator system to realize fault location.Without collecting offline data to generate a sample set,the online fault feature data set is directly used in this method,which is generated by the satellite attitude control system to perform fault detection.And fault of the satellite attitude control system is effectively located according to fault detection results.Simulation results demonstrate the effectiveness of the proposed method.