Abstract:The optimization and selection of temperature measuring points is a difficulty in the thermal error compensation technology for numerical control (NC) machine tool. The temperature field of a vertical milling machine was obtained through the thermal imager, according to the distribution of temperature field, some temperature sensors were installed on the machine tool. Using FCM fuzzy clustering method and the correlation analysis, the temperature measuring points were grouped and optimized on the basis of the measurement data of the temperature and thermal deformation, and then the thermal error model of the key temperature measuring point was established by using multiple regressions analysis, and verified by the experiment. The results show that this method can effectively reduce the measuring points, temperature measuring point is reduced from 13 to 5, and the precision of built model is better. Y, Z direction of the thermal error is reduced from 50 μm to 9 μm.