Machining accuracy reliability optimization of three-axis CNC machine tools using doubly-weighted vector projection response surface method

Abstract

The reasonable allocation of geometric errors of machine tools can improve their machining accuracy reliability (MAR). However, due to the complexity and high nonlinearity of limit state function (LSF) of MAR, the fitting accuracy is usually low when the traditional method is used to approximate LSF. To solve this problem, a doubly-weighted vector projection response surface (DWVPRS) method, which considers not only the approximation results of test sample points (TSPs) to LSF but the distances between TSPs and the most probable failure point (MPFP), is proposed. Using the reliability sensitivity analysis method, the key geometric errors were identified and optimized. Finally, taking a large gantry guideway grinding machine as an example to verifies the effectiveness and correctness of the DWVPRS method proposed in this paper, the results show that compared with the traditional methods, the DWVPRS method has the highest fitting accuracy to approximate LSF at the MPFP, and after the optimization of geometric accuracy, both the minimum and average reliability values of the grinding machine meet the design requirements.