Abstract
Abstract
The non-contact R-test is used for the dynamic measurement of five-axis machine tools. However, due to spherical curvature, calculating the spherical center displacement (SCD) using the non-contact R-test is challenging and inefficient. Thus, in this paper, an efficient calculation method based on multiple Gaussian process regression (GPR) is proposed to calculate the SCDs of the continuous trajectory by using the non-contact R-test. In the proposed method, first, the transformation matrix is constructed to roughly calculate the SCD. Second, the measurement error caused by the spherical curvature is solved and compensated by using the GPR model. Next, the threshold is set to remove some of the set points and reconstruct GPR to improve the model accuracy. Finally, the transformation matrix and the GPR model are outputted to calculate the SCDs of the continuous trajectory. We constructed a micro S trajectory to verify the proposed method. Experimental results showed that compared with the neural network method, the proposed method achieves higher model accuracy with shorter model construction time and improves the calculation efficiency by 99%.
Funder
National Natural Science Foundation of China
Subject
Applied Mathematics,Instrumentation,Engineering (miscellaneous)
Cited by
1 articles.
订阅此论文施引文献
订阅此论文施引文献,注册后可以免费订阅5篇论文的施引文献,订阅后可以查看论文全部施引文献