Prediction of surface residual stress in end milling with Gaussian process regression

The residual stress has an important influence on the performance of parts such as fatigue life, and many researches have been carried out for the quantitative evaluation or prediction of residual stress. However, the quantitative evaluation of residual stress achieved by the instrument will cause damage to the parts or consume a long time. Moreover, some analytical models for machining-induced residual stress prediction are quite complex with poor universality. In this research, surface residual stress under different cutting parameters and machining features were investigated. The optimal feature set and feature importance were analyzed by random forest algorithm. A new method for predicting machining-induced surface residual stress was proposed based on Gaussian process regression and other machine learning algorithms were also utilized to verify the performance of the proposed approach. The values of quantitative evaluation indicators showed that the prediction model provides a robust framework for surface residual stress prediction.