This paper presents results from experimental and numerical studies of the skew rolling process used to shape axisymmetric products made of C60-grade steel. An experimental study was carried out to investigate the effect of process parameters described by the forming angle α, the skew angle θ, the reduction ratio δ, and the jaw chuck velocity Vu on the surface roughness Ra of the forgings. Stepped forgings made of C60-grade steel were rolled. Based on numerical calculations, a machine learning regression model was developed that uses process parameters to predict the surface roughness of produced parts. The random forest model was found to be the most effective based on the determined metrics (MAE, RMSE, R2). A more detailed analysis of this model was performed using the SHAP library. The application of ML methods will enable optimization of skew rolling through appropriate selection of process parameters affecting improvement in product quality.
Keywords: C60 steel; CNC rolling mill; machine learning models; metal forming; roughness; skew rolling; stepped axles and shafts.