Research On Thermal Error Compensation Techniques Of CNC Machine Tools

Thermal error of CNC machine tools is the major contributor of total errors. Error compensation technique is one of the most effective methods to improve the accuracy of machine tools. This dissertation is focused on the technique of thermal error compensation. Based on the research of global study in this field, some main problems were proposed in this dissertation. The study in this dissertation is focused on the selection of thermal key points, thermal characteristic analysis of machine tool spindle, thermal error modeling of machine tool spindle and the verification experiment. The work has been completed as follows:Firstly, based on the theory of finite element method and calculating the heat transfer convection coefficient of a flat plate when air flows along it, new methods of calculating the heat transfer convection coefficient of the rotating spindle surface and other boundary coefficients were proposed. The dynamic thermal characteristic of machine tool spindle and the relationship between temperature of key points and thermal errors were analyzed by using finite element analysis program ANSYS and numerical calculation program MATLAB.Secondly, based on the theory of thermal sensitivity and practical experience, a new method of distinguishing thermal key points was proposed. Base on the hysteresis and nonlinear characteristic of the spindle and Elman network theory, a modified Elman network was proposed to establish thermal error model of machine tool spindle. The accuracy of the model was tested by using the data from finite element analysis of the spindle, and the result was good.Finally, based on the theory of virtual instrument and Lab VIEW, the verification experiment and data collection software were established. The accuracy of the thermal error model was tested by using the data from verification experiment. And the result demonstrated that the thermal error model has high level of robustness capability, and could reduce thermal error by more than 80 percent. The modified Elman network is more suitable to establish thermal error model of the spindle than Elman network.