A new modeling method for thermal errors of motorized spindle based on the variation characteristics of spindle temperature field

Motorized spindle is a component with complex thermal properties. Its thermal boundary conditions change with operating conditions, and thus, the variations of spindle temperature field under different conditions are not exactly the same. Models with fixed structures and fixed coefficients have difficulty in predicting the spindle thermal error with high accuracy and stability. To solve this problem, study of the variation rules of the spindle temperature field is conducted in this paper. The variation process is divided into two stages: the non-uniform variation stage and the uniform variation stage. The variation characteristics are concluded into two parts: the overall variation tendency and the variation of temperature distribution from the inside to outside of the spindle. These conclusions provide a new perspective for temperature variable selection and coefficients adjustment. Based on the variation characteristics of the spindle temperature field, a new modeling method for spindle thermal errors is proposed. The basic temperature, the difference between temperature near the internal heat source and the basic temperature, and the ambient temperature are applied as input variables. The coefficient of the second input variable is set to adjust adaptively according to the spindle status. This method puts more emphasis on the variation rules of the spindle temperature field and can significantly increase the accuracy and the robustness of the prediction model. The model coefficients are optimized by PSO (particle swarm optimization). According to the experimental verification, the variation ranges of residual errors are reduced by at least 62% when predicting the thermal error in X-direction with the proposed model.