Title of article :
ResidualstresspredictionofdissimilarmetalsweldingatNPPsusingsupportvectorregressionOriginalResearchArticle
Author/Authors :
ManGyunNa، نويسنده , , JinWeonKim، نويسنده , , DongHyukLim، نويسنده , , Young-JuneKang، نويسنده ,
Issue Information :
روزنامه با شماره پیاپی سال 2008
Abstract :
Residualstressesareanimportantfactorinthecomponentintegrityandlifeassessmentofweldedstructures.Inthispaper,asupportvectorregression(SVR)methodispresentedtopredicttheresidualstressfordissimilarmetalweldingaccordingtovariousweldingconditions.DissimilarweldingjointbetweennozzleandpipeisregardedintheanalysessinceithasbeenknowntobehighlysusceptibletoPrimaryWaterStressCorrosionCracking(PWSCC)intheprimarysystemofanuclearpowerplant(NPP).Theresidualstressdistributionsarepredictedalongtwostraightpathsofaweldzone:apipeflowpathontheinnerweldsurfaceandapathconnectingtwocentersoftheinnerandoutersurfacesofaweldzoneofapipe.FourSVRmodelsaredevelopedforfournumericaldatagroupswhicharesplitaccordingtothetwoendsectionconstraintsandthetwopredictionpathsandtheSVRmodelsareoptimizedbyageneticalgorithm.TheSVRmodelsaretrainedbyusingadatasetpreparedfortraining,optimizedbyusinganoptimizationdataset,andverifiedbyusingatestdatasetindependentofthetrainingdataandtheoptimizationdata.ItisknownthattheSVRmodelsaresufficientlyaccuratetobeusedintheintegrityevaluationbypredictingtheresidualstressofdissimilarmetalweldingzones.
Journal title :
Nuclear Engineering and Design Eslah
Journal title :
Nuclear Engineering and Design Eslah
