简介：Aso-called“localprobabilisticParisrelationmethod”waspresentedformeasuringtherandomthresholdsoflongfatiguecrackpropagation.Acheckwasmadetotheconventionalmethod,inwhichthethresholdsweremeasuredstatisticallyanddirectlybythetestdata.Itwasrevealedthatthismethodwasnotreasonablebecausethetestdatahaveseldomaunifiedlevelofcrackgrowthrates.Differently,inthepresentedmethodtheParis-Erdoganequationwasappliedtomodelthelocaltestdataaroundthethresholds.Localprobabilisticrelationswithboththesurvivalprobabilityandtheconfidencewereestablishedonalognormaldistributionofthestressdensityfactors.Andthen,theprobabilisticthresholdswerederivedfromtheprobabilisticfactorswithagivencriticallevelofgrowthrate.AnanalysisonthetestdataofLZ50axlesteelfortheChineserailwayvehiclesverifiesthatthepresentmethodisfeasibleandavailable.