Abstract: Theattitudekinematicequationsandmeasurementequationsarethefunctionoftheorbitparameters,sothestandardExtendedKalmanFilter(EKF)can′tgiveoptimalattitudeestimationbecauseoftheorbiterrors,especiallyforamagnetometerbasedsmallsatellite.ArobustadaptiveKalmanfilterwasdesignedtodealwiththedisturbancefromtheorbitestimationerrors.Theprincipleofrobustestimationwasadoptedfortheadaptivefilters,andthecovariancematrixandobservationalnoisewereadjustedtogetproperKalmangains,andthentheapproximatelyoptimalfilteringresultswereobtained.BasedonEKFstabletheory,RAKFwasproveduniformlyandstochasticallystableifthesystemisasymptoticallycontrollableandasymptoticallyobservable.SimulationresultsindicatethatRAKFcanimproveattitudeestimationprecisionfrom0.3°to0.2°,andthefilterisapprovedtobemoreeffectiveandrobustforattitudeestimation.

Key words: Magnetometer, Attitudeestimation, Robustadaptivefilter, Kalmangain, Stability, Smallsatellite