节点文献

基于自编码器与softmax的ACS敏感器并发故障定位

ACS Sensor Concurrent Faults Detection and Location Based on Autoencoder and Softmax

  • 推荐 CAJ下载
  • PDF下载
  • 不支持迅雷等下载工具,请取消加速工具后下载。

【作者】 程月华; 叶正宇; 姜斌; 韩笑冬;

【Author】 CHENG Yuehua;YE Zhengyu;JIANG Bin;HAN Xiaodong;College of Automation Engineering,Nanjing University of Aeronautics and Astronautics;China Academy of Space Technology;

【机构】 南京航空航天大学自动化学院; 中国空间技术研究院;

【摘要】 航天器姿态敏感器冗余配置提升了系统可靠性和可重构性,却增加了部件故障发生概率,对敏感器并发故障的辨识和定位重要而具有挑战性。本文提出一种自编码器与Softmax组合应用的敏感器并发故障定位方法,分为4步:残差生成,特征提取,故障辨识与故障定位。首先根据航天器姿态测量原理设计基于自编码器的观测器,比较观测器与敏感器输出生成残差。然后提取残差特征。再采用Softmax多分类器对提取的特征信号进行辨识以辨识故障。最后通过多观测器表决的方法定位故障。仿真结果表明所提方法能实现姿控系统敏感器的并发故障辨识与定位。

【Abstract】 The redundant configuration of spacecraft attitude sensors improved the reliability and reconfigurability of the attitude control system(ACS), while increased the probability of component fault. It is of great significance but challenging to realize concurrent faults identification and location. We propose a sensor concurrent fault location method based on autoencoder and softmax, which is divided into 4 steps, residuals generation, features extraction, fault identification and fault location. In the step of residuals generation, autoencoder based observers are designed based on the spacecraft attitude determination principles, residual signals are generated by comparing the output of observers and sensors. Then extract features of residual. And softmax multi-classifiers are applied to identify the extracted features to identify the faults. Finally, the faults are located by using the method of multiple observer voting. The experiment results show that the proposed method is capable of realizing concurrent fault identification and location of sensors of ACS.

【基金】 国家自然科学基金(61972398);十三五装备预先研究项目(30501050403);基本科研业务费(NC2020002,NP2020103);高等学校学科创新引智计划资助(B20007)
  • 【会议录名称】 2020中国自动化大会(CAC2020)论文集
  • 【会议名称】2020中国自动化大会(CAC2020)
  • 【会议时间】2020-11-06
  • 【会议地点】中国上海
  • 【分类号】V467
  • 【主办单位】中国自动化学会
节点文献中: 

本文链接的文献网络图示:

本文的引文网络