Research on electric spindle thermal error prediction model based on DBO-SVM

This article proposes an improved method to address the issue of low precision in the conventional thermal error projection model for electric spindles. Firstly, thermal error experiments were conducted on the electric spindles under different operating conditions to collect data on the temperature and axial-radial displacement offsets at measuring points. The variations of temperature and displacement over time were analyzed. Furthermore, the LAFCM clustering and grey correlation analysis were employed to identify the three optimal temperature measurement points from a total of ten measurement points. Subsequently, an improved thermal error prediction model was constructed using the optimized temperature variables as inputs and the axial thermal error as the output. This model combined the dung beetle optimizer (DBO) algorithm and support vector machines (SVM), with the DBO algorithm optimizing the SVM parameters. The resulting model demonstrated higher prediction accuracy, robustness, and generalization ability. This method provides a theoretical basis and technical support for compensating and optimizing the thermal error of electric spindles.