To predict the geometrical characteristics and cross-sectional average hardness of composite cladding layer of U71Mn steel laser cladding WC-Ni60, and to provide reference for repairing rail wear by laser cladding technology. Different experiments were designed and single-track WC-Ni60 composite coatings were deposited on the surface of U71Mn steel by later cladding technique.Meanwhile,the cross-sectional macrostructure and average hardness of the single-track coating was observed by industrial microscopy and Leitzs microhardness tester. Then BP, GA-BP and PSO-BP network were used to predict, and the result of them were compared with the experimental results. The prediction accuracy of GA-BP and PSO-BP model was both good,the average relative errors of the GA-BP model and the PSO-BP model were 2.53% and 2.90%, respectively. The GA-BP model was more accurate in predicting the average hardness of the cladding layer, the average relative errors were 1.00%. The PSO-BP model was more stable than the GA-BP model.So the accuracy of GA-BP model was higher, and the stability of PSO-BP model was better.