Study on Fault Detection of Star-Sensor Based on UKF

Di Hu,Dan Liu,Guangming Bai

doi:10.1051/jnwpu/20193761158

Abstract

The fault detection of star sensor in satellite when no telemetry data from the gyroscope can be obtained is investigated. An algorithm based on the unscented Kalman filter (UKF) is proposed to indicate the occurrence of the fault. By adding the angular speed of the satellite into the state equation, the UKF is designed to estimate the angular and the angular speed of the satellite. After that, a comparison between the measured and the estimated value of both the angular and the angular speed is made. In this way, the star sensor fault can be finally detected. Two fault cases (abrupt constant fault and noise increase fault) are taken into consideration in the simulation to show the validness of the present algorithm. Furthermore, a test verification by using the real telemetry data on orbit is performed to demonstrate that the present algorithm can detect the star sensor fault effectively. In application, the developed fault detection algorithm can be employed in the ground measurement and control station to monitor the star sensor fault such that the fault can be detected immediately and the alarm indicating the occurrence of the fault will be given.

Highlights

An algorithm based on the unscented Kalman filter
into consideration in the simulation to show the validness of the present algorithm
the de⁃ veloped fault detection algorithm can be employed in the ground measurement and control station to monitor the star sensor fault such that the fault can be detected

Summary

Introduction

摘要:针对通信卫星在轨运行后无陀螺遥测数据的星敏故障问题,提出一种基于无迹卡尔曼滤波 (unscented kalman filter,UKF)的故障检测算法,将卫星角速度纳入状态方程进行估计,采用 2 个星敏的输出数据作为测量方程,通过分别比对正常时段和故障时段下卫星角度的稳定值与其估计值、角速度的稳定值与其估计值,可以准确检测出星敏数据故障。针对在轨可能出现的 2 种故障工况(突变故障、噪声变大故障) 进行了仿真,并进一步采用在轨遥测数据进行了试验验证,仿真和试验验证结果均表明,该算法能有效检测出星敏数据故障,可用于地面对在轨星敏数据进行故障监测,以便发现故障后及时进行报警处理。近年来,随着技术的进步,在轨运行的通信卫星越来越多,由于通信卫星具有高可靠、长寿命的特点,如何在星上进行故障检测并实施系统重构,确保卫星在轨运行时的业务不受影响已成为目前广大研究人员关注焦点。此外,针对寿命末期的卫星,如何在地面对星上单机进行有效监测,以确保故障能够被及时处理,也成为目前必须解决的问题。卫星控制分系统是卫星系统的重要组成部分,它直接决定了卫星在轨运行的姿态情况,以及卫星业务是否正常运行。姿态敏感器作为控制分系统的测量元件, 一旦发生故障,将会给卫星带来灾难性的后果。由于当前在轨运行的大部分通信卫星,其陀螺为非长寿命陀螺,无法长期开机,因此针对在轨无陀螺的通信卫星,对其星敏感器进行故障检测算法研究是非常必要的。给出了陀螺故障检测算法,是通过对陀螺的漂移估计实现的,因此陀螺也必须开机。文献[ 12] 采用自适应 UKF 对敏感器故障设计了故障检测算法,但该算法成立的条件是陀螺和星敏感器均开机。针对星敏感器故障的情况,文献[ 13] 利用陀螺与星敏的冗余信息,提出了一种线性最小均方差估计的故障诊断算法。本文在文献[10⁃11] 的基础上,研究了只有姿态角测量信息时的星敏感器故障检测问题。通过将卫星角速度纳入状态方程, 并设计 UKF 进行状态估计,以获得角度和角速度的残差信息,进而通过合并检测,并将其与阈值比较,即可检测出星敏感器的故障。最后以 2 种常见的星敏故障为例进行仿真,并进一步采用在轨数据对所提算法进行试验验证,仿真和试验验证结果均表明,本文设计的算法可以有效地对星敏感器进行故障检测。

Results

Conclusion