基于BP神经网络的数控机床误差辨识方法研究  

Study on NC Machine Error Identification Based on BP Neural Network

作　　者：李耀明[1] 沈兴全[1] 孟庆义 王爱玲[1] 

机构地区：[1]中北大学机械工程与自动化学院,山西太原030051 [2]北京卫星制造厂,北京100190

出　　处：《中北大学学报（自然科学版）》2009年第6期574-578,共5页Journal of North University of China(Natural Science Edition)

基　　金：山西省青年自然科学基金资助项目(2009021022-2);山西省研究生创新基金资助项目

摘　　要：数控机床误差辨识是进行误差补偿的关键技术,一般很难通过建立准确的数学模型对误差进行辨识.而神经网络具有良好的非线性函数逼近能力,因此可将其应用于数控机床误差辨识.研究了基于BP神经网络的数控机床测头测量误差参数的辨识方法.通过理论分析和实验研究,可确定测头安装误差参数和测头测量误差参数的模型,还可确定消除测头系统不确定性误差项的条件.研究为确定检测方法和规划检测路径提供了可靠的依据.NC machine＇s error identification is a key technology for error compensation.It＇s difficult to identify the error by establishing an accurate mathematical model.The neural network has a good nonlinear function approximation capability,which can be used in NC machine error identification.Parameter identification method of NC machine＇s tool probe measurement error was proposed based on BP neural network.Through theoretical analysis and experimental research,the models of the probe＇s installation and measurement error parameters,and the condition to eliminate uncertainty of probe system error term can be determined.The research can provide reliable basis for determining the detection method and planning test path.

关 键 词：BP神经网络 数控机床 辨识 

分 类 号：TH16[机械工程—机械制造及自动化]