点击次数：

论文类型： 期刊论文

发表时间： 2018-01-01

发表刊物： PRECISION ENGINEERING-JOURNAL OF THE INTERNATIONAL SOCIETIES FOR PRECISION ENGINEERING AND NANOTECHNOLOGY

收录刊物： SCIE、EI

卷号： 51

页面范围： 208-222

ISSN号： 0141-6359

关键字： Binocular vision; Vision measurement; Position independent geometric errors; Five-axis machine tool; Error identification; Image analysis

摘要： Errors of the rotary axes are the main sources of geometric errors in a five-axis machine tool. Thus, accurate periodic checks and calibrations of the rotary axes are important for improving the machining precision. In this paper, to achieve error detection with three-dimensional (3D) measurement capability and to simplify the complex error identification formulations of position-independent geometric errors (PIGEs) in the rotary axis (C axis), both a binocular-vision-based error detection system and an identification algorithm are proposed. First, the 3D error detection system is investigated, in which a novel self-luminous cooperative target is designed to characterize the movement information of a rotary table; thus, high-precision installation and high-signal-tonoise image acquisition are realized. In addition, to further guarantee the 3D vision measurement accuracy, a center location method based on reconciled conjugate constraints is adopted to improve the position accuracy of the rotary table. Then, in the error identification process, an error identification model that is independent of the machine structure is established to separate the error parameters, which simplifies the complex mathematical formulations. By best-fitting a set of 3D measurement positions to the identification algorithms, each error parameter of the PIGEs can be separately identified by simply mounting the cooperative target on the rotary table in one setup. Experiments for the measurement and identification of PIGEs in C-axis of a five-axis machine tool were performed in a laboratory; in comparison with the identification results obtained by double-ball bar (DBB) test, the experimental results verified the vision-based error identification accuracy and feasibility.