摘要
Inthenormaloperationcondition,aconventionalsquare-rootcubatureKalmanfilter(SRCKF)givessufficientlygoodestimationresults.However,ifthemeasurementsarenotreliable,theSRCKFmaygiveinaccurateresultsanddivergesbytime.ThisstudyintroducesanadaptiveSRCKFalgorithmwiththefiltergaincorrectionforthecaseofmeasurementmalfunctions.Byproposingaswitchingcriterion,anoptimalfilterisselectedfromtheadaptiveandconventionalSRCKFaccordingtothemeasurementquality.Asubsystemsoftfaultdetectionalgorithmisbuiltwiththefilterresidual.Utilizingaclearsubsystemfaultcoefficient,thefaultysubsystemisisolatedasaresultofthesystemreconstruction.Inordertoimprovetheperformanceofthemulti-sensorsystem,ahybridfusionalgorithmispresentedbasedontheadaptiveSRCKF.Thestateanderrorcovariancematrixarealsopredictedbythepriorifusionestimates,andareupdatedbythepredictedandestimatedinformationofsubsystems.Theproposedalgorithmswereappliedtothevesseldynamicpositioningsystemsimulation.TheywerecomparedwithnormalSRCKFandlocalestimationweightedfusionalgorithm.ThesimulationresultsshowthatthepresentedadaptiveSRCKFimprovestherobustnessofsubsystemfiltering,andthehybridfusionalgorithmhasthebetterperformance.Thesimulationverifiestheeffectivenessoftheproposedalgorithms.
出版日期
2013年01月11日（中国Betway体育网页登陆平台首次上网日期，不代表论文的发表时间）
