Thermal error modeling of motorized spindle and application of miniature radiator in motorized spindle

The precision of machine tools heavily relies on the motorized spindle’s performance. It is crucial to guarantee the thermal error prediction model’s improved accuracy, as it enables enhanced machining accuracy, reduced errors, and effective control over the prediction model. The northern goshawk optimization (NGO) algorithm is used to optimize the kernel extreme learning machine (KELM) to address thermal error issues. Based on the analysis conducted, the best four temperature measurement locations are chosen. The established thermal error prediction model, combining NGO and KELM, is compared with both the basic KELM model and BA-KELM through comprehensive analysis. The results show that the determination coefficient R2 of NGO-KELM model is 0.132 and 0.026 higher than KELM and BA-KELM. The robustness, stability, and generalization ability of the NGO-KELM modeling method are verified. In order to further reduce the thermal error, this paper puts forward a method of adding a microradiator to the contact between the front and rear bearing chambers and the bearings. The findings demonstrate that it is possible to efficiently lower the temperature at the front and rear bearings and suppress the thermal deformation of the motorized spindle.