Prediction Model for Surface Roughness of Polycarbonate Using Single-Point-Diamond-Turning Lathe Machining Based on Machine Learning Techniques

The goal of this research is to use Machine Learning (ML) models to forecast surface roughness in the manufacture of Polycarbonate (PC) using the Single-Point-Diamond-Turning (SPDT) process. Feed rate, cut depth, X-, Y-, and Z-axis vibrations, and spindle speed are the predictors of the SPDT process of ultra-precision turning. In this research, four regression approaches were used to predict surface roughness: linear regression (LIN), support vector regression (SVR), gradient boosting regression (GBR), and random forest regression (RFR) (denoted by Ra). The error metrics root-mean-squared-error (RMSE), mean absolute-error (MAE), and coefficient of determination (R-2) were used to evaluate the predictive performance of prediction models. The GridSearchCV algorithm was used to find the optimum hyperparameters in order to improve the prediction ability of each model. The results showed that the SVR model performed best, with the lowest RMSE and MAE and the highest R-2. This suggests that SVR was the most accurate model for predicting PC surface roughness using the SPDT procedure.