粒子滤波的原理和实现-渥太华大学课件-theory-and-implementation-of-particle-filters.ppt

TheoryandImplementationofParticleFiltersMiodragBolicAssistantProfessorSchoolofInformationTechnologyandEngineeringUniversityofOttawa12Nov20041BigpictureGoal:EstimateastochasticprocessgivensomenoisyobservationsConcepts:BayesianfilteringMonteCarlosamplingsensortObservedsignal1tObservedsignal2ParticleFiltertEstimation12Nov20042ParticlefilteringoperationsParticlefilterisatechniqueforimplementingrecursiveBayesianfilterbyMonteCarlosamplingTheidea:plexproblemscontinuesLargenumberofapplicationsrequireevaluationofintegralsNon-linearmodelsNon-Gaussiannoise12Nov20045SequentialMonteCarloTechniquesBootstrapfilteringThecondensationalgorithmParticlefilteringInteractingparticleapproximationsSurvivalofthefittest12Nov20046HistoryFirstattempts–,“MonteCarlocalculationoftheaverageextensionofmolecularchains,”JournalofChemicalPhysics,,,–359,-,,,“Novelapproachtononlinear/non-GaussianBayesianstateestimation,”IEEProceedings-F,,,–113,,,,Eds.,SequentialMonteCarloMethodsinPractice,Springer,,,,BeyondtheKalmanFilter:ParticleFiltersforTrackingApplications,ArtechHousePublishers,,,,,“Atutorialonparticlefiltersforonlinenonlinear/non-gaussianBayesiantracking,”IEEETransactionsonSignalProcessing,,,–188,worksOtherapplications1)Biology&BiochemistryChemistryEconomics&BusinessGeosciencesImmunologyMaterialsSciencePharmacology&ToxicologyPsychiatry/,,,"OnSequentialMonteCarloSamplingMethodsforBayesianFiltering", puting,,,-208