Error Prediction of CMM Using a Hybrid Model Based on Neural Network Quantile Regression and Kernel Density Estimation

Abstract

Abstract The sources of dynamic measurement error of CMM are complex and influence each other and the traditional parameter modeling method is very difficult to model. In this paper, we propose a hybrid model which combines neural network quantile regression and kernel density estimation. The hybrid model realizes the advantages of multi-angle analysis, nonlinear fitting data and non-parametric error prediction. We use this model to analyze the complex relationship between dynamic measurement error value and 3d coordinates, positioning velocity, proximity distance and contact velocity. The results show that our model has good predictive performance and is superior to the least squares estimation model.