Predicting residual properties of ball screw raceway in whirling milling based on machine learning

The residual properties of the ball screw raceway after whirling milling are the critical factors affecting the performance of the workpiece. Comprehensive research has been conducted to investigate the residual properties (including the residual stress and full-width half-maximum (FWHM)) of the ball screw raceway in dry machining. More importantly, two kinds of machine learning methods including the neural network (NN) and support vector machine (SVM) have been proposed based on experimental data to predict the residual properties of the ball screw raceway. In the method of NN, two representative methods including the back-propagation neural network (BPNN) and radial basis function neural network (RBFNN) have been employed. The model verifications show that the average absolute relative errors of the residual stress perpendicular to the cutting direction and FWHM are very small by using the method of machine learning. The comparative analysis of the predictive models shows that SVM has a better prediction performance than NN. The accurate prediction of residual properties by SVM can provide technical support for the optimization of machining parameters and improve the performance of ball screw.