The application discloses a kind of
laser cladding repair axle class part fatigue life prediction method and
system, comprising: obtaining each test piece microdefect feature, the heterogeneous structure feature of the
microstructure of cladding layer-
heat affected zone-substrate, carry out rotating
bending fatigue test, obtain the rotating
bending fatigue life data corresponding to microdefect feature, heterogeneous structure feature, and establish the fatigue life
label corresponding to data;Establish the initial
data set including cladding layer microdefect
feature parameter, heterogeneous structure
feature parameter and axle fatigue life data;And the
data set is pretreated and reduced dimension
processing;Establish
feature extraction model, adopt
random forest regression model to carry out feature importance quantification, identify key feature;The fatigue life prediction model of XGBoost
algorithm is constructed, the
feature parameter of the high independence of microdefect and heterogeneous structure is obtained, and the key feature parameter and weight are screened out, to carry out residual fatigue life prediction based on
fracture mechanics framework.