Abstract:Due to the problems of temperature-sensitive point selection and model establishment in the modeling of CNC machine tools thermal error compensation, the method was presented by combined with fuzzy clustering and grey correlation to select temperature-sensitive points and the autoregressive distributed lag was used to establish model. According to the experimental data of machine temperature and thermal error, multiple regression model and autoregressive distributed lag model were built respectively. The modeling test of thermal error was designed on the Leaderway V-450 CNC machining center, the thermal error and temperature data were measured on the conditions of the spindle speed in 2000, 4000 and 6000r/min. The result showed that fitting accuracy of distributed lag mode was better than that of multiple regression model, robustness of distributed lag mode was lower than that of multiple regression model when experimental data of any spindle speed was used to modeling, but the robustness of distributed lag mode was prior to multiple regression model when experimental data of any two spindle speeds were used to modeling. Application of autoregressive distributed lag model for CNC machine tools thermal error prediction can be useful.