融合时间的单温度敏感点机床热误差建模方法  

A time-integrated approach to modeling thermal errors in machine tools with a single temperature-sensitive point

作　　者：唐光元 苗恩铭 王文辉 石照耀[2] TANG Guangyuan;MIAO Enming;WANG Wenhui;SHI Zhaoyao(College of Mechanical Engineering,Chongqing University of Technology,Chongqing 400054,China;Beijing Research Center for Precision Measurement and Control Technology and Instrumentation Engineering Technology,Beijing University of Technology,Beijing 100124,China)

机构地区：[1]重庆理工大学机械工程学院,重庆400054 [2]北京工业大学北京市精密测控技术与仪器工程技术研究中心,北京100124

出　　处：《重庆理工大学学报（自然科学）》2023年第11期300-307,共8页Journal of Chongqing University of Technology：Natural Science

基　　金：国家重点研发计划项目(2019YFB1703700);重庆市技术创新与应用发展专项重点项目(cstc2019jscx-mbdxX0045,cstc2019jscx-mbdxX0016)。

摘　　要：针对敏感点变动性和敏感点共线性对热误差预测模型的预测精度和稳健性的影响,提出融合时间的单温度敏感点建模方法,将隐性参数时间显性化,进一步明确温度变量、时间变量与热误差之间的关系,提升热误差模型的预测精度和稳健性,降低温度敏感点的选择难度。在采用半年数控机床热误差实验数据时,仅选用一个温度测点作为敏感点,建立温度、时间与热误差之间的多元回归预测模型;与传统的选择多温度敏感点的多元回归热误差预测模型进行比对分析,验证所提方法的有效性。研究结果表明:对数控机床的Y向热误差,所提出建模方法的平均预测精度为2.57μm,模型稳健性为1.37μm,相较于传统的热误差预测模型,预测精度和稳健性提高了28.0%和47.1%;对数控机床的Z向热误差,所提出建模方法的平均预测精度为5.30μm,模型稳健性为3.40μm,相较于传统的热误差预测模型,预测精度和稳健性提高了45.1%和57.7%;能较好地降低温度敏感点的选择难度,提高热误差模型的预测精度和稳健性。The variability of the sensitive point and the covariance of the sensitive point affect the prediction accuracy and robustness of the thermal error prediction model.This paper proposes a time-integrated single temperature sensitive point modeling method,making the implicit parameter time explicit and further clarifying the relationship between the temperature variable,the time variable and the thermal error.The method improves the prediction accuracy and robustness of the thermal error model,and makes it easier to select the temperature sensitive point.From the thermal error data of CNC machine tools collected from a six-month-long experiment,the proposed method chooses only one temperature measurement point as the sensitive point and builds a multivariate regression prediction model between temperature,time and thermal error.A comparison is conducted with the traditional multiple regression thermal error prediction model that has multiple temperature sensitive points to determine the effectiveness of the method.The results show for the Y-direction thermal error of a CNC machine tool,the proposed modeling method has an average prediction accuracy of 2.57μm and a model robustness of 1.37μm,which are 28.0%and 47.1%higher than those of the traditional thermal error prediction model;for the Z-direction thermal error of a CNC machine tool,the proposed modeling method has an average prediction accuracy of 5.30μm and a model robustness of 3.40μm,which are 45.1%and 57.7%higher than those of the traditional thermal error prediction model.The proposed thermal error modeling method better overcomes the difficulties in selecting the temperature sensitive points and improves the prediction accuracy and robustness of the thermal error model.

关 键 词：数控机床 热误差 时间 单温度敏感点 多元线性回归 

分 类 号：TH161[机械工程—机械制造及自动化]