Surface Hardness Intelligent Prediction in Milling Using Support Vector Regression

Abstract

Surface hardness is a major factor affecting the performance of a component. Understanding the influence of machining parameters on surface hardness is very important for the control of workpiece quality. In milling process development, it is highly desirable to predict the surface hardness of machined workpiece. For this purpose, a novel prediction model based on support vector regression (SVR) is developed to investigate the influence of processing parameters on the milled surface hardness. Firstly, the influence of spindle speed, feed rate, cutting depth and milling cutter on the milled surface hardness (from experimental data) are taken into account. Then the experimental procedure employed a full factorial design with these foresaid parameters as factors, each of first three terms at three levels and the last at two levels, are carried out and 54 groups of data are obtained. Furthermore, the data are analyzed by different experiments in contrast: SVR based models and BP based model respectively. The experimental results have indicated that SVR based model outperforms its main competitors-BP model in the limited training data, and the mean absolute error of which is only 20%-30% of the latter. Lastly, This prediction model provides a better understanding of the influence of milling conditions on machined surface hardness. In sum, the proposed model is faster in speed, higher in accuracy, and more suitable for prediction of the machined surface hardness.