Research on thermal error of electric spindle based on ISCA-GRNN

Abstract

To meet the requirement for enhancing the machining precision of electric spindles in high-speed precision machine tools, we have optimized and designed a new type of electric spindle, selecting a cage-type asynchronous AC motor. Thermal properties have been improved by optimizing the integration of heat sources and the cooling system to tackle heat accumulation in bearing components. Experimentation confirmed that, at speeds ranging from 6,000 to 10,000 revolutions per minute (rpm), the new spindle exhibited a 19.61% decrease in axial thermal displacement and a 59.52% reduction in radial thermal displacement compared to conventional spindles. In order to further control the thermal error accurately, a new type of thermal error prediction model for electric spindles is constructed by integrating the Improved Sine Cosine Algorithm (ISCA) and Generalized Regression Neural Network (GRNN). Using the experimental data in the above speed intervals, the thermal errors of the new electrospindle are predicted, and compared with the unimproved SCA-GRNN prediction model, the ISCA-GRNN prediction model reduces the mean absolute error by 38.90% and the root-mean-square error by 27.47%, which is a significant performance enhancement that verifies the validity of the proposed model.