三坐标测量机动态误差混合建模方法

Hybrid modeling method for CMM dynamic error

为了建立高精度的三坐标测量机相关性动态误差修正模型,应用偏最小二乘回归法和支持向量机研究动态误差混合建模方法,分析了三坐标测量机的动态误差来源及特点,介绍了偏最小二乘回归法和支持向量机的基本原理及算法,编制了相应的混合建模软件。应用双频激光干涉仪设计了测量机动态误差实验装置,进行了相应的误差分离实验。应用偏最小二乘法确定测量机动态误差主要影响因素,得出坐标变量y、z和x是主要影响因素、测量速度v是次要影响因素的结论。结合主要影响因素分析结果,应用支持向量机和动态误差实验数据建立了测量机的空间动态误差模型,并进行比对和验证,应用混合建模方法建立的修正模型精度可以达到±0.02μm,高于未利用偏最小二乘法±0.75μm的建模精度,完全可以用于三坐标测量机相关性动态误差的高精度建模。

In order to build a high-precision CMM correlated dynamic error correction model,the partial least-squares regression method （PLSRM） and new support vector machine method （nu-SVM） are used to study the hybrid dynamic error modeling method.The dynamic error characteristics and sources of CMM are analyzed in detail.The principle and algorithm of the partial least-squares regression method and new support vector method are introduced.Corresponding hybrid modeling software was programmed.A CMM dynamic error separation experiment facility was designed using laser interferometer and corresponding experiments were carried out.The data were analyzed using PLSRM and the conclusions are the coordinates y and z and x are the major error influence factors and the measuring velocity v is the minor factor.A CMM spatial dynamic error model was built using the nu-SVM and dynamic error experiment data.The model was verified,and the conclusion is that the modeling precision reaches ±0.02 μm,which is better than the ±0.75 μm modeling precision for the model without adopting PLSRM.The built CMM dynamic error model can be used to correct the CMM strong correlation dynamic error with high precision.