Predictive modeling of surface roughness in lenses precision turning using regression and support vector machines

Abstract

Slow tool servo (STS) turning is superior in machining precision and in complicated surface. However, STS turning is a complex process in which many variables can affect the desired results. This paper focuses on surface roughness prediction in lenses STS turning. An exponential model, based on the five main cutting parameters including tool nose radius, feed rate, depth of cut, C-axis speed, and discretization angle, for surface roughness prediction of lenses is developed by means of orthogonal experiment regression analysis. Meanwhile, a prediction model of surface roughness based on least squares support vector machines (LS-SVM) with radial basis function is constructed. Orthogonal experiment swatches are studied, and chaotic particle swarm optimization and leave-one-out cross-validation are applied to determine the model parameters. The comparison of LS-SVM model and exponential model is also carried out. Predictive LS-SVM model is found to be capable of better predictions for surface roughness and has absolute fraction of variance R 2 of 0.99887, the mean absolute percent error e M of 8.96 %, and the root mean square error e R of 10.68 %. The experimental results and prediction of LS-SVM model show that effects of tool nose radius and feed rate are more significant than that of depth of cut on surface roughness of lenses turning.