简介:Inthenormaloperationcondition,aconventionalsquare-rootcubatureKalmanfilter(SRCKF)givessufficientlygoodestimationresults.However,ifthemeasurementsarenotreliable,theSRCKFmaygiveinaccurateresultsanddivergesbytime.ThisstudyintroducesanadaptiveSRCKFalgorithmwiththefiltergaincorrectionforthecaseofmeasurementmalfunctions.Byproposingaswitchingcriterion,anoptimalfilterisselectedfromtheadaptiveandconventionalSRCKFaccordingtothemeasurementquality.Asubsystemsoftfaultdetectionalgorithmisbuiltwiththefilterresidual.Utilizingaclearsubsystemfaultcoefficient,thefaultysubsystemisisolatedasaresultofthesystemreconstruction.Inordertoimprovetheperformanceofthemulti-sensorsystem,ahybridfusionalgorithmispresentedbasedontheadaptiveSRCKF.Thestateanderrorcovariancematrixarealsopredictedbythepriorifusionestimates,andareupdatedbythepredictedandestimatedinformationofsubsystems.Theproposedalgorithmswereappliedtothevesseldynamicpositioningsystemsimulation.TheywerecomparedwithnormalSRCKFandlocalestimationweightedfusionalgorithm.ThesimulationresultsshowthatthepresentedadaptiveSRCKFimprovestherobustnessofsubsystemfiltering,andthehybridfusionalgorithmhasthebetterperformance.Thesimulationverifiestheeffectivenessoftheproposedalgorithms.