Thermal error modeling of electrical spindle based on optimized ELM with marine predator algorithm

Abstract

High-speed electric spindle is an important part of computer numerical control (CNC) machining equipment, and the thermal displacement generated by the electric spindle during operation affects the electric spindle machining stability and machining accuracy. Error compensation for the high-speed motorized spindle can well compensate for the influence of thermal displacement on machine tool processing. Therefore, the prediction of thermal displacement of electric spindle is particularly important. In this paper, firstly, the temperature field information is obtained by simulating and analyzing the calculation of the heat generation and heat exchange inside the electric spindle. Set up the experimental platform with reasonable temperature measurement points according to the simulation results; secondly, the temperature sensitive points are screened by fuzzy C-mean clustering algorithm and Pearson correlation coefficient, which can effectively improve the covariance and correlation between temperature variables; finally, based on the screened data for thermal error modeling, an optimized extreme learning machine based on marine predator algorithm (MPA-ELM) is provided to predict the thermal displacement of electric spindles model. And comparing the model accuracy of extreme learning machine (ELM), MPA-ELM and extreme learning machine optimized by genetic algorithm (GA-ELM), the experimental data show that MPA-ELM has better prediction accuracy.