The thermal characteristics of a motorized spindle are the main determinants of its performance, and influence the machining accuracy of computer numerical control machine tools. It is important to accurately predict the thermal field of a motorized spindle during its operation to improve its thermal characteristics. This paper proposes a model to predict the temperature field of a high-speed and high-precision motorized spindle under different working conditions using a finite element model and test data. The finite element model considers the influence of the parameters of the cooling system and the lubrication system, and that of environmental conditions on the coefficient of heat transfer based on test data for the surface temperature of the motorized spindle. A genetic algorithm is used to optimize the coefficient of heat transfer of the spindle, and its temperature field is predicted using a three-dimensional model that employs this optimal coefficient. A prediction model of the 170MD30 temperature field of the motorized spindle is created and simulation data for the temperature field are compared with the test data. The results show that when the speed of the spindle is 10,000 rpm, the relative mean prediction error is 1.5%, and when its speed is 15,000 rpm, the prediction error is 3.6%. Therefore, the proposed prediction model can predict the temperature field of the motorized spindle with high accuracy.

Aiming at the problem that changes of nonlinear dynamic resistance of stator affect the performance of speed sensorless vector control system, a hybrid computing intelligence approach is used in the identification of stator resistance of motorized spindle. The partial least squares (PLS) regression is combined with neural network to solve the problem of few samples and multi-correlation of variables in complicated data modeling. The PLS method is used to extract variable components from sample data and then reduced the dimension of input variables. Moreover, neural network is used to fit the non-linearity between input and output variables. The model based on partial least squares regression and neural network can identify stator resistance under different conditions of the motorized spindle. The results show that the method has high identification precision and is helpful to improve the performance of vector control system.