Spatial Measurement Error Estimation of Coordinate Measuring Machine Based on Kernel Function

Due to the complicated system constitution of three-dimensional measuring machine, it involves the multiple error components. So in the fast measurement, it is complicated to analyze spatial error characteristics in detail. In the experimental planning section, data are collected by measuring Gauge block in different length and different space positions. Then we use partial least square and kernel function to model and predict the experimental data. According to the result, the error of mean square in kernel function is 0.00542mm and the correlative coefficient is as high as 91.1373% while the mean square of partial least square is 0.3069mm. The experiment results indicate that the kernel function has a better imitative predictive effect than the partial least square and non-parametric modeling is more accurate than the parametric modeling in predicting spatial measurement error estimation of Coordinate Measuring Machine.