›› 2012, Vol. 32 ›› Issue (6): 23-30.doi: 10.3780/j.issn.1000-758X.2012.06.004

• 研究探讨 • 上一篇    下一篇

磁定姿近地轨道卫星EKF与UKF算法比较

 杨斌1, 徐广涵1,2, 靳瑾3, 周游1   

  1. (1北京航空航天大学电子信息工程学院,北京100191)(2北京信威通信技术股份有限公司,北京100094)(3清华大学航天航空学院,北京100084)
  • 收稿日期:2012-02-23 修回日期:2012-04-19 出版日期:2012-12-25 发布日期:2012-12-25
  • 作者简介:杨 斌 1982年生,2007年获中北大学信号与信息处理专业硕士学位,现为北京航空航天大学通信与信息系统专业博士研究生。主要研究方向为自适应滤波。

ComparisononEKFandUKFforGeomagneticAttitudeEstimationofLEOSatellites

 YANG  Bin1, XU  Guang-Han1,2, JIN  Jin3, ZHOU  You1   

  1. (1SchoolofElectronicandInformationEngineering,BeihangUniversity,Beijing100191)
    (2BeijingXinweiTelecomTechnologyInc.,Beijing100094)
    (3SchoolofAerospace,TsinghuaUniversity,Beijing100084)
  • Received:2012-02-23 Revised:2012-04-19 Online:2012-12-25 Published:2012-12-25

摘要: 某近地轨道(LEO)卫星通过磁强计测量、飞轮控制和磁力矩器卸载,实现三轴姿态稳定。围绕该卫星姿态确定问题,对扩展卡尔曼滤波(EKF)和无迹卡尔曼滤波(UKF)算法进行综合分析对比,最后给出该卫星姿态确定实现方法。分析了地磁模型误差、磁强计测量误差及剩磁的影响,推导了四元数均值及协方差的计算方法。分别引入EKF和UKF进行姿态确定,并讨论了影响算法性能的因素。对两种方法的估计精度、收敛时间及计算需求等指标进行了对比。最后针对该卫星姿态确定要求和硬件条件,选用EKF算法实现姿态确定,满足对地指向精度优于1°,稳定度优于0.1(°)/s的设计指标。

关键词: 扩展卡尔曼滤波, 无迹卡尔曼滤波, 地磁测量, 姿态确定, 低轨卫星

Abstract: AlowEarthorbit(LEO)satellitecanbestabilized    in3-axisbycorporatelyusingamagnetometer,reactionwheelsand3-axismagnetorquerrods.Itischallengingtodeterminate anappropriateattitudeforLEO.Toaddressthischallenge,thealgorithmsofextendedKalmanfilter(EKF)andunscentedKalmanfilter(UKF)werecomparedandasolutionwasproposedbyexploringtheinfluenceofthegeomagneticmodelerror,themagneticmeasurementerrorandtheremanenceontheperformanceofattitudedetermination.Inaddition,thecalculationofmeanandcovariancequaternionwasderived.TheEKFandUKFwereintroducedtoattitudedetermination,andthefactorswhichhaveinfluenceonthealgorithmperformancewerediscussed.Twoalgorithmswerecomparedonestimationaccuracy,convergencetimeandcomputationcomplexity.Withtheaccuracyrequirementsandthehardwarelimitations,theEKFalgorithmwasfinallychosenfortheattitudedetermination.Simulationresultsshowthattheearthpointingaccuracyoftheproposedsolutioncanreach1°withthestabilityof0.1(°)/s.

Key words: Kalmanfilter, UnscentedKalmanfilter, Geomagneticmeasurement, Attitudedetermination, LowEarthorbitsatellite