宇航学报

宇航学报 ›› 2021, Vol. 42 ›› Issue (4): 513-521.doi: 10.3873/j.issn.1000-1328.2021.04.012

• 制导、导航、控制与电子 • 上一篇    下一篇

利用随机森林算法的卫星控制系统故障诊

黄瑾，刘洋，钟麦英，杨瑞，李文博，刘成瑞   

  1. 1. 山东科技大学电气与自动化工程学院，青岛 266590；2. 西交利物浦大学智能工程学院，苏州 215123；3. 北京控制工程研究所，北京 100190；4. 空间智能控制技术重点实验室，北京 100190

  • 收稿日期:2020-11-18 修回日期:2021-01-21 出版日期:2021-04-15 发布日期:2021-04-15
  • 基金资助:
    国家自然科学基金（61733009，61873149，61703244）；空间智能控制技术实验室开放基金课题（HTKJ2019KL502007）；山东省泰山学者特聘教授人才团队资助项目

Fault Diagnosis of Satellite Attitude Control Systems Using Random Forest Algorithm

HUANG Jin, LIU Yang, ZHONG Mai ying, YANG Rui, LI Wen bo, LIU Cheng rui   

  1. 1. College of Electrical Engineering and Automation, Shandong University of Science and Technology, Qingdao 266590, China;2. School of Advanced Technology, Xi’an Jiaotong Liverpool University, Suzhou 215123, China;3. Beijing Institute of Control Engineering, Beijing 100190, China; 4. Science and Technology on Space Intelligent Control Laboratory, Beijing 100190, China
  • Received:2020-11-18 Revised:2021-01-21 Online:2021-04-15 Published:2021-04-15

PDF

180

摘要： 针对卫星姿态控制系统存在闭环控制，外部干扰强，进而影响故障诊断准确性和实时性的问题，提出了一种利用随机森林算法的卫星姿态控制系统姿态敏感器和执行器故障诊断方法。首先，采集不同故障情形下卫星姿态控制系统的输入输出数据，进行特征提取。随后通过随机有放回抽样划分训练集和测试集，建立基于随机森林算法的故障诊断模型。利用生成的随机森林模型，对实时输入输出数据进行分类，实现卫星姿态控制系统的故障诊断。利用卫星姿态控制系统半物理仿真平台的数据进行实验，对比实验结果表明：本文所提方法可以实现故障高精度分离，并具有更好的实时性，适用于卫星姿态控制系统的故障诊断问题。

关键词: 卫星姿态控制, 故障诊断, 随机森林, 决策树

Abstract: To cope with the closed loop controls and strong external disturbances, a fault diagnosis method based on the random forest (RF) algorithm is proposed for satellite attitude control systems. Firstly, various types of signals in satellite attitude control systems in different faulty scenarios are collected. Then, some characteristics of the data are extracted and a RF based fault diagnosis model is established correspondingly. The possible faults can be diagnosed via injecting the real time data into the obtained RF based model. The effectiveness of the proposed method is illustrated on a semi physical simulation platform, and the comparison results show that our approach can identify a variety of faults with high accuracy and efficiency.

Key words: Satellite attitude control, Fault diagnosis, Random forest, Decision trees

中图分类号: