摘要
为建立一种能够适应机床不同工况且具有准确预测能力的热误差补偿模型,提出一种基于限定记忆递推最小二乘法辨识热误差模型参数的机床热误差预测建模方法。该方法随着机床工作状况的改变,根据实时反馈的温度和热误差数据,采用递推方法对模型参数进行即时修正,使热误差模型能够及时跟踪机床系统的热特性变化,实现以较高的预测精度对机床热误差进行补偿。通过数控车床主轴轴向热误差辨识建模及补偿实验可以看出,限定记忆递推最小二乘法比一步最小二乘法辨识精度有较大提高,最大残差值减小了52.3%,标准差减小了67%。实验结果表明,利用该方法进行机床热误差模型参数辨识具有较高的预测精度和鲁棒性,有效可行。
In order to compensate the thermal errors with accurate prediction under various working conditions of machine tools, a novel thermal error prediction modeling method with RFMLS method was presented. As variations in working condition of machine tools, according to real-time online feedback data of temperature and thermal deformation of key measuring points, thermal error model could trace the thermal characteristics by rameters timely, so the thermal compensation using recursive algorithm to update the model＇s pawould be carried out successfully. The method was verified by an axial thermalerror compensation experiment of spindle conducted on a CNC lathe. The experimental results show that predictive model with RFMLS is more accurate than that with least squares method, the maximum residual error and standard deviation of the former decrease by 52.3% and 67% than those of the latter respectively. Hence the thermal error of machine tools compensated by the presented method can improve accuracy and reduce the model predictive errors effectively.
出处
《中国机械工程》
EI
CAS
CSCD
北大核心
2015年第3期361-365,共5页
China Mechanical Engineering
基金
国家自然科学基金资助项目(51375382)
国家科技重大专项(2012ZX04012032)
陕西省自然科学基础研究计划重点项目(S2009JC1400)
关键词
数控机床
热误差
在线补偿
限定记忆递推最小二乘法
CNC machine tool
thermal error
on-line compensation
finite memory recursive leastsquares（RFMLS）
