引用本文:叶正宇,程月华,韩笑冬,姜斌.深空探测航天器姿态控制系统故障定位[J].控制理论与应用,2019,36(12):2093~2099.[点击复制]
YE Zheng-yu,CHENG Yue-hua,HAN Xiao-dong,JIANG Bin.Fault location for attitude control systems of deep space exploration satellites[J].Control Theory and Technology,2019,36(12):2093~2099.[点击复制]
深空探测航天器姿态控制系统故障定位
Fault location for attitude control systems of deep space exploration satellites
摘要点击 1926  全文点击 768  投稿时间：2019-06-22  修订日期：2019-11-02
查看全文  查看/发表评论  下载PDF阅读器
DOI编号  10.7641/CTA.2019.90468
  2019,36(12):2093-2099
中文关键词  故障定位  神经网络  支持向量机  姿态控制系统
英文关键词  Fault location  Neural networks  Support vector machines  Attitude control system
基金项目  装备预研国防科技重点实验室基金(1422080307). 国家自然科学基金(61972398). 十三五装备预先研究项目(30501050403). 研究生创新基地（实验室）开放基金(kfjj20180321).
作者单位E-mail
叶正宇 南京航空航天大学 kasoll076@outlook.com 
程月华* 南京航空航天大学 chengyuehua@nuaa.edu.cn 
韩笑冬 通信卫星事业部  
姜斌 南京航空航天大学  
中文摘要
      执行机构与敏感器故障检测与定位是深空探测任务卫星平台可靠运行的前提和保障。本文从数据的角度出发，结合姿控系统工作机理，提出一种基于神经网络和支持向量机结合的故障诊断方法用于检测并定位故障。故障诊断方法分为3步，首先采集姿控系统的状态信息，采用神经网络对闭环姿控系统中未知动态特性建模并进行预测；然后将姿控系统敏感器信号与神经网络预测输出比较生成残差并提取故障特征；最后采用支持向量机辨识残差特征检测故障，并结合运动学特性分析定位故障。仿真结果表明本文所提方法可以有效提取、辨识故障特征，实现执行器与敏感器的故障检测定位。
英文摘要
      Fault detection and location is the prerequisite and guarantee for the reliability of the satellite for deep space exploration missions. Combined with the analysis of the attitude control system (ACS) mechanism, we present a data-driven fault diagnosis (FD) method based on neural networks and support vector machines (SVM) to detect and locate fault. The FD scheme is composed of 3 steps. Firstly, the sensor signals of the ACS are collected, and a neural network is applied to model the unknown dynamic characteristic of closed-loop ACS and make predictions; Then, ACS sensor signals are compared with the neural network predictions to generate residuals and extract the fault features; Finally, an SVM is applied to identify the fault features to realize fault detection. And combined with kinematic analysis, fault location is realized. The simulation results show that the proposed method can effectively extract and identify the fault features and is capable of realizing the fault detection and location of actuator and sensor.