Machine Learning approach to the prediction of Surface roughness of turned Glass/Basalt epoxy composites

Abstract

In this work, the Machine Learning techniques namely Support Vector Regression, Random forest method and Extreme Gradient Boosting  (XGBOOST) are utilized for the prediction of Surface Roughness in the turning process of Glass/Basalt epoxy hybrid composites. The experiments were conducted in accordance with the Taguchi's L27 orthogonal array. The experimental results indicates that, the surface roughness of the turned Glass/Basalt epoxy composites decreases with the increase in Spindle speed, decrease in Feed rate and Depth of cut. It was also observed that feed rate has a greatest impact and Depth of cut has a least effect over the surface roughness while the spindle speed moderately influenced the surface roughness. From the results of Machine Learning models, it is evident that the Random forest model  appears to be superior with a Mean Absolute error and Maximum error of 4.96% and 7.73% respectively for testing data set.