Robustness Analysis of the Thermal Error Model for a CNC Machine Tool

Abstract

To compensate the thermal error of a CNC machine tool with the thermal error prediction model is an effective and efficient method to reduce thermal errors. The robustness of the thermal error prediction model has significant influence on the consequence of thermal error compensation. The multivariate linear regression and RBF neural network are usually used to build thermal prediction model. In this paper these two methods are used to build thermal prediction model of a CNC machining center CR5116, and the robustness analysis of the two models are also made. The experimental validation on the CNC machining center CR5116 shows that the accuracy of the prediction models is affected by some key temperature measuring points. Hence the robustness of the thermal error prediction models is improved through adding the redundant key temperature measuring points.