Intelligent Modeling and Predicting Surface Roughness in End Milling

Predicting the effects of manufacturing conditions on surface roughness is very important for the control of work-piece quality. In this study least square-support vector regression (LS-SVR) is used for predicting surface roughness of end milling surface with related to cutting parameters and phenomena. Three cutting parameters (spindle speed, feed rate, depth of cut), and vibrations as input vector and corresponding surface roughness of work-pieces as output result were firstly collected for training and testing data set. On the basis of training data set, three prediction models for surface roughness using back-propagation (BP) neural network, standard support vector regression (SVR), and LS-SVR are developed, respectively. Accuracies of those models are tested on the testing data set. The LS-SVR based model is found to be superior over the others in terms of training speed and accuracy for the prediction. The results lead to a good understanding of the influence of milling conditions on surface roughness.