Surface roughness optimal estimation for disc parts turning using Gaussian-process-based Bayesian combined model

Abstract

The surface roughness is an important characterization of the products performance, and its estimation is required to evaluate the machining accuracy level of the turning machining or its machining condition monitoring. Traditional methods are using machining parameters or combined machining parameters with tool vibration to predict the surface roughness. But for a steel disc part turning machining, the surface roughness value on a circumference trajectory is not the same as one section of the trajectory because randomness element will exist in surface roughness value. This paper proposed a machine learning approach of using Gaussian-process-based Bayesian combined model to construct surface roughness trajectory prediction. Gaussian-process-based Bayesian combined model is a good machine learning method for dealing with variable with random element and introduced into building the correlation between surface roughness and surface roughness increment for next point in a trajectory. To show the process of the method establishing the model, a simulation is carried on first. Then, a turning experiment was conducted. The experimental results verify that Gaussian-process-based Bayesian combined model compared with Gaussian-process model can be used to predict the surface roughness in a more reliable way.

Recent Advances on Machine Learning Applications in Machining Processes

“…Liu et al [66] considered Gaussian modeling and Bayesian learning to implement a surface roughness estimation in ring-shaped thin-walled discs to reduce the number of measurements required based on measured values along different trajectories. Another Bayesian application [22] applied as an input the vibration and process parameter time domain features and the radial basis function-based kernel principal component analysis (KPCA-IRBF) as a feature selection to apply the Sparse Bayesian Linear Regression (SBLR) for the roughness prediction, obtaining a Root Mean Square Error (RMSE) of 0.0317 and a Pearson Correlation Coefficient of 0.9926.…”