Research on thermal error modeling and prediction of heavy CNC machine tools

Abstract

Thermal error has been a significant factor influencing the accuracy of heavy CNC machine tools. A thermal experiment is performed on a typical heavy CNC machine tool. According to the temperature field of the machine tool, a new hierarchical clustering method is proposed to optimize temperature variables efficiently. Euler distance and correlation coefficient are both considered in the method, by which the collinearity of temperature variables is reduced. Thermal error prediction model is built based on optimized temperature variables utilizing MRA. Results show that the RMSE of thermal error prediction can be reduced to lower than 10 μm, which has higher accuracy compared to other methods. This method is expected to be used in thermal error modeling and prediction of other heavy CNC machine tools.