Modeling of NC Machine Tool Thermal Error Based on Adaptive Best-fitting WLS-SVM

Abstract

In order to eliminate the influence of thermal error of NC machine on the machining precision, a novel method based on adaptive best-fitting weighted least squares support vector machine (WLS-SVM) is utilized to implement error compensation. In order to construct the thermal error model of machine tool, a series of experiments are carried out to acquire the data of a XK713 NC milling machine, including temperature on different positions and the thermal deformation of spindle. By smart temperature sensors and laser position sensors, the temperature and thermal error of the machine tool are collected respectively. First the parameters of WLS-SVM are optimized by a method called adaptive best-fitting parameter search algorithm. Then the samples are trained and the weighted coefficients are calculated according to the error variables. The regression model is constructed finally. Test results show that WLS-SVM is an effective method for error modeling which can be used for the thermal error compensation. It is superior to unweighted least squares support vector machine (LS-SVM) method and traditional least squares (LS) method.