Surface Roughness Analysis and Prediction with an Artificial Neural Network Model for Dry Milling of Co–Cr Biomedical Alloys

The aim of this paper is to conduct an experimental study in order to obtain a roughness (Ra) prediction model for dry end-milling (with an AlTiCrSiN PVD-coated tool) of the Co–28Cr–6Mo and Co–20Cr–15W–10Ni biomedical alloys, a model that can contribute to more quickly obtaining the desired surface quality and shortening the manufacturing process time. An experimental plan based on the central composite design method was adopted to determine the influence of the axial depth of cut, feed per tooth and cutting speed process parameters (input variables) on the Ra surface roughness (response variable) which was recorded after machining for both alloys. To develop the prediction models, statistical techniques were used first and three prediction equations were obtained for each alloy, the best results being achieved using response surface methodology. However, for obtaining a higher accuracy of prediction, ANN models were developed with the help of an application made in LabView for roughness (Ra) prediction. The primary results of this research consist of the Co–28Cr–6Mo and Co–20Cr–15W–10Ni prediction models and the developed application. The modeling results show that the ANN model can predict the surface roughness with high accuracy for the considered Co–Cr alloys.