摘要
Amethodwasdevelopedtorealizequalityevaluationoneveryweld-spotinresistancespotweldingbasedoninformationprocessingofartificialintelligent.Firstly,thesignalsofweldingcurrentandweldingvoltage,asinformationsource,weresynchronouslycollected.Inputpoweranddynamicresistancewereselectedasmonitoringwaveforms.Eightcharacteristicparametersrelatingtoweldqualitywereextractedfromthemonitoringwaveforms.Secondly,tensile-shearstrengthofthespot-weldedjointwasemployedasevaluatingtargetofweldquality.Throughcorrelationanalysisbetweeneverytwoparametersofcharacteristicvector,fivecharacteristicparameterswerereasonablyselectedtofoundamappingmodelofweldqualityestimation.Atlast,themodelwasrealizedbymeansofthealgorithmsofRadialBasicFunctionneuralnetworkandsamplematrixes.Theresultsshowedvalidationsbyasatisfactioninevaluatingweldqualityofmildsteeljointon-lineinspotweldingprocess.
出版日期
2008年04月14日（中国期刊网平台首次上网日期，不代表论文的发表时间）
