Predictive Modeling of Thermal Displacement for High-Speed Electric Spindle

Abstract

Accurate, efficient and stable prediction of thermal displacements generated during spindle machining is essential for improving machining quality, increasing economic efficiency and ensuring production safety. Aiming at the existing thermal displacement prediction models with low precision and poor robustness, this paper put forward a prediction model based on the Bald Eagle Search (BES) algorithm optimized Least Squares Support Vector Machine (LSSVM). Firstly, the experimental platform was built to carry out the spindle thermal deformation experiment and collect the experimental data. Then use K-means clustering method to classify the temperature measurement points, and combine with gray correlation analysis to calculate the size of the correlation between each point and thermal displacement, comprehensive analysis of the classification results and the size of the correlation, from the 10 points preferred 4 points. After that, the BES algorithm, which has strong searching ability in the global range, is chosen to optimize the internal parameters of LSSVM, and the prediction model based on BES-LSSVM is constructed by learning the nonlinear correlation characteristics between the spindle temperature and axial thermal displacement. Finally, it is compared with the prediction model using BES algorithm to optimize support vector machine and the prediction model using sparrow search algorithm to optimize LSSVM respectively. The comparison reveals that the predictions output from the BES-LSSVM model have better accuracy and stability. The results of the study can provide a certain knowledge base and technical support for the effective prediction of spindle thermal displacement changes.