A method for predicting the morphology of single-track laser cladding layer based on SO-LSSVR

Abstract

The morphology of the cladding layer is directly affected by the process parameters of laser cladding. To improve the morphology of single-track cladding layer, it is necessary to accurately reveal the nonlinear relationship between the morphology and process parameters of laser cladding. Firstly, the single-track Ni60/10%WC cladding layer was preparate on the surface of 35CrMoV substrate, and the effects law of process parameters such as laser power, scanning speed and powder feed rate on the morphology of single-track cladding layers were investigated through single factor experiment. Then, the prediction model for predicting the morphology of single-track laser cladding layer was established by the SO-LSSVR. At the same, the prediction models based on ternary quadratic regression, SVR, and LSSVR were established for comparative analysis. Finally, the prediction results show that the MAPE, RMSE and R2 of the SO-LSSVR prediction model are better than the prediction results of the remaining three models. The proposed SO-LSSVR prediction model can accurately describe the nonlinear relationship between the process parameters and morphology of single-track cladding layer, which lays a foundation for the subsequent process parameter optimization to improve the morphology of cladding layer.