Surface Roughness Prediction with Machine Learning

Abstract

Surface roughness is an exceedingly essential index to measure the quality of materials. For manufacturing and production, it is very significant yet difficult to precisely measure the surface roughness of materials. Based on input features including the rational speed, feed and depth of cutting, we managed to predict the surface roughness with machine learning models, based on a real-world dataset collected in the laboratory. The process and results are discussed in this paper. We demonstrate that the neural network can achieve a better prediction performance, compared with other baseline models. This insight will be of great help to the manufacture industry.