加工中心主轴关键热敏感点选取与热误差预测

Thermal Error Prediction of Machine Tool Spindle Based on GA-ACO-BP Network

为探究数控机床主轴温度场信息与主轴热误差之间的非线性映射关系,提出一种基于人工蜂群优化算法(ABC)与广义回归神经网络的主轴热误差预测模型。首先,使用热成像技术布置温度传感器,并利用K-medoids算法对温度测点进行聚类分组,使用灰色关联度分析方法计算温度与主轴热误差之间的相关程度,进而提取出最佳热敏感点;其次,引入调节因子优化ABC算法的寻优过程,使用改进后的ABC网络确定GRNN网络的最佳参数及光滑因子;最后,以三轴数控加工中心为研究对象,进行温度数据与热误差数据的采集,建立基于ABC-GRNN热误差预测模型并与优化前进行比较。热误差实验结果表明,K-medoids算法与灰色关联分析相结合,有效避免了温度测点之间的多重共线性;ABC-GRNN模型可以更准确地预测出主轴各项误差值。

To investigate the nonlinear mapping relationship between the spindle temperature field information of CNC machine and the thermal error of the spindle,A spindle thermal error prediction model based on artificial bee colony optimization algorithm(ABC) and generalized regression neural network was proposed.Firstly,thermal imaging technology is used to arrange temperature sensors,and K-medoids algorithm is used to cluster temperature measurement points,Grey correlation degree analysis was used to calculate the correlation degree between temperature and thermal error of spindle,which in turn extracts the optimal thermal sensitivity point.Secondly,Introduction of a regulating factor to optimize the optimization process of the ABC algorithm,and the optimal parameters and smoothness factor of GRNN network are determined by using the improved ABC network.Finally,the three-axis CNC machining center was taken as the research object,the temperature data and thermal error data were collected,and the thermal error prediction model based on ABC-GRNN was established and compared with before optimization.The experimental results of thermal error show that the K-medoids algorithm combined with grey correlation analysis can effectively avoid the multicollinearity between temperature measurement points,The ABC-GRNN model can predict the each error values of the main axis more accurately.