A new thermal error modeling method for CNC machine tools

A great challenge in a thermal error compensation process is to select proper temperature variables and to establish accurate thermal error models. In this paper, a new approach for building an effective mathematic thermal error for machine tools is presented. Fuzzy clustering analysis is conducted to identify temperature variables, and then screen the representative variable as an independent variable meanwhile eliminate the coupling among the variables. Cluster validity is forwarded to measure the reasonability of the clustering and the classification accuracy for the temperature variables. Furthermore, a mathematical model using the robust regression analysis is built to reveal the relationship between these temperature variables and thermal deformation. To evaluate the performance of our proposed model, a verification experiment is carried out. Pt-100 thermal resistances and Eddy current sensors are used to monitor the temperature and thermal shift fluctuation respectively. Fuzzy clustering analysis is utilized to classify 32 temperature variables to four clusters. A robust regression thermal error model is proposed based on the four key temperature points. The result shows that four representative temperature variables are precise predictors of the thermal errors of the machine tools. The proposed method is shown to be capable of improving the accuracy of the machine tools effectively.