Thermal error modeling of high-speed electric spindle based on Aquila Optimizer optimized least squares support vector machine

Abstract

The thermal error of electric spindle is an important factor affecting the machining accuracy of machine tools. To establish the thermal error prediction model, the thermal characteristics of A02 electric spindle were simulated and analyzed by ANSYS, to determine the best temperature measuring point. The temperature and thermal error of A02 electric spindle at different rotational speeds were obtained by thermal characteristics experiment. Through the combination of systematic clustering and grey relational analysis, four best temperature measuring points were selected from ten temperature measuring points. A prediction model of least squares support vector machine (LSSVM) optimized by Aquila Optimizer (AO) is established and compared with the model optimized by Particle Swarm Optimization (PSO), it is superior to the latter in prediction accuracy (η), goodness of fit (R2), root mean square error (RMSE) and mean absolute error (MAE). The AO-LSSVM prediction model based on experimental data has a prediction accuracy of more than 94% for the thermal error of motorized spindle and has good stability and generalization ability.

Keywords

• Aquila optimizer
• Least squares support vector machine
• Electro-spindle
• System clustering
• Grey relational degree