Application of information fusion to volumetric error modeling of CNC machine tools

Abstract

In order to solve the problem of low measurement efficiency of volumetric errors and predict the volumetric errors of CNC machine tools online, a new volumetric error modeling method based on information fusion technology is proposed in this paper, which was used for the volumetric error prediction of a two turntable five-axis machine tool. In order to fulfill the volumetric error modeling, the displacement variables, temperature variables, and cutting force variables were determined firstly, then the determined variables and the volumetric errors were measured and analyzed, and a volumetric error model was presented based on the determined variables. In order to optimize the parameters of the presented model, grey correlation analysis method is used for the second time prediction of the volumetric errors. Finally, the performance of the model was tested by an experiment. The result shows that the approximation ability of the model is very well, and the residual error is smaller than 2 μm. Comparing with conditional modeling methods, this method can be used in different types of machine tools with different operating conditions by regulating the weight matrix, and the robustness of the fusion model is improved.

Thermal Error Prediction and Compensation of YK3610 Hobbing Machine Based on BP Neural Networks

“…In order to improve the machining accuracy of gear hobbing machine, a lot of studies focused on thermal error prediction and compensation has been presented recently, such as artificial neural networks [8,9], GA-BPN method [10], project pursuit regression method [11], information fusion method [12], and so on. In this study, BP neural network algorithm was applied to predict thermal errors of a YK3610 hobbing machine, and ant colony algorithm was *Address correspondence to this author at the School of Mechanical Engineering, Shandong University of Technology, Zibo 255049, China; Tel: +86-2781723; E-mail: guoqianjian@163.com used to train the link weights of BP neural network model, which overcomes the local minimum problem of BP neural networks and improves the prediction performance of thermal error modeling.…”