数控机床热误差分组建模优化研究  

Research on Optimized Group-Wise Modeling of Thermal Errors in a CNC Machine Tool

作　　者：赵海涛 雷鸣[1] 谌海莲[1] 凌晓辉[1] ZHAO Haitao;LEI Ming;CHEN Hailian;LING Xiaohui(Physical Science and Technology College,YiChun University,Yichun 336000,China)

机构地区：[1]宜春学院物理科学与工程技术学院,江西宜春336000

出　　处：《机械》2025年第2期8-14,58,共8页Machinery

基　　金：江西省教育厅科技项目(190837)。

摘　　要：对于加工参数突变的较长加工过程,传统多变量回归热误差模型会产生局部较差的预测精度。为此,提出分组建模以改善预测精度。首先借助硬分断点和软分断点设计对温度变量、热误差变量采样数据序列分组的方法;其次,以热误差拟合残差和为目标函数,利用遗传算法实现硬分断点的优化选择;第三,给出组间热误差模型系数更替的判别依据。最后在一台车削中心上针对径向热误差进行建模验证,验证数据来自有限元仿真。从热误差预测结果看,误差预测精度随着硬分断点数的增加而增加,但增加幅度越来越小,且硬关键点位置具有相对稳定性;当分断点数从0增加至4时,径向热误差拟合残差和从0.1153 mm减少到0.0331 mm。因此,分组热误差模型对提高预测精度是有效的。For a long machining process with abrupt changes of machining parameters,the conventional multivariate regression analysis based thermal error model can produce a local poorer prediction accuracy,so a group-wise modeling method is proposed in this paper.Firstly resort to hard break points and soft break points,design a method for grouping of the sampling data series for temperature variables and thermal error variables.Secondly realize the optimized selection of hard break points using genetic algorithm with the sum of fitting residuals of thermal errors as the objective function.Thirdly provide a method for updating of coefficient vectors for different soft groups.The validation test results for the radial thermal error for the given working conditions on a turning center show that the prediction accuracy increase with the more number of the hard break points,but the increasing magnitude becomes smaller,furthermore the locations of the hard break points are relatively fixed;the fitting residual sum for the radial thermal errors decreases from 0.1153 mm to 0.0331 mm when the number of hard break points is from 0 to 4,which means the group-wise thermal error model is helpful to improve the predicting accuracy.

关 键 词：热误差 数控机床 优化分组 建模 遗传算法 

分 类 号：TH161[机械工程—机械制造及自动化]