Thesis - Stochastic Modeling of Deterioration in Nuclear Pow

《Thesis - Stochastic Modeling of Deterioration in Nuclear Pow》由会员上传分享，免费在线阅读，更多相关内容在学术论文-天天文库。

1、StochasticModelingofDeteriorationinNuclearPowerPlantComponentsbyXianxunYuanAthesispresentedtotheUniversityofWaterlooinful¯lmentofthethesisrequirementforthedegreeofDoctorofPhilosophyinCivilEngineeringWaterloo,Ontario,Canada,2007°cXianxunYuan2007IherebydeclarethatIamthesoleauthorofth

2、isthesis.Thisisatruecopyofthethesis,includinganyrequired¯nalrevisions,asacceptedbymyexaminers.Iunderstandthatmythesismaybemadeelectronicallyavailabletothepublic.iiAbstractTherisk-basedlife-cyclemanagementofengineeringsystemsinanuclearpowerplantisintendedtoensuresafeandeconomicallye

3、±cientoperationofenergygenerationin-frastructureoveritsentireservicelife.Animportantelementoflife-cyclemanagementistounderstand,modelandforecastthee®ectofvariousdegradationmechanismsa®ectingtheperformanceofengineeringsystems,structuresandcomponents.Themodelingofdegradationinnuclear

4、plantcomponentsisconfoundedbylargesamplingandtemporaluncertainties.Thereasonisthatnuclearsystemsarenotreadilyaccessibleforinspectionsduetohighlevelofradiationandlargecostsassociatedwithremotedatacollectionmethods.Themodelsofdegradationusedbyindustryarelargelyderivedfromordinaryline

5、arregressionmethods.Themainobjectiveofthisthesisistodevelopmoreadvancedtechniquesbasedonstochasticprocesstheorytomodeldeteriorationinengineeringcomponentswiththepurposeofprovidingmorescienti¯cbasistolife-cyclemanagementofagingnuclearpowerplants.Thisthesisproposesastochasticgammapro

6、cess(GP)modelfordeteriorationanddevelopsasuiteofstatisticaltechniquesforcalibratingthemodelparameters.Thegammaprocessisaversatileandmathematicallytractablestochasticmodelforawidevarietyofdegradationphenomena,andanotherdesirablepropertyisitsnonnegative,monotonicallyincreasingsamplep

7、aths.Inthethesis,theGPmodelisextendedbyincludingadditionalcovariatesandalsomodelingforrandome®ects.Theoptimizationofage-basedreplacementandcondition-basedmaintenancestrategiesisalsopresented.Thethesisalsoinvestigatesimprovedregressiontechniquesformodelingdeterioration.Alinearmixed-

8、e®ects(LME)regressionmodel