Identification of position-dependent geometric errors with non-integer exponents for Linear axis using double ball bar. (15th March 2020)

• Record Type: Journal Article
• Authors: Xu, Kai; Li, Guolong; He, Kun; Tao, Xiaohui
• Is Part Of: International journal of mechanical sciences. Volume 170(2020)
• Publication Date: 2020-03-15
• Subjects: Error identification -- Double ball bar -- Position-dependent geometric error -- Non-integer exponents; Mechanical engineering -- Periodicals; Génie mécanique -- Périodiques; Mechanical engineering; Maschinenbau; Mechanik; Zeitschrift; Periodicals; 621.05
• DOI: 10.1016/j.ijmecsci.2019.105326
• Languages: English
• ISSNs: 0020-7403

Abstract

Highlights: The value of position dependent geometric errors varies as the position changes. Pre-fitted method indirectly identifies errors by solving coefficients of polynomial. Non-integer exponents fitting can achieve a comparable accuracy as polynomial. Non-integer exponents can avoid linear correlation problem in pre-fitted method. Modification with constant is necessary in error identification in 3 planes. Abstract: Geometric accuracy is significant for machine tools. Identification of geometric errors, especially position-dependent geometric errors (PDGEs) is difficult because of the limitations of many factors such as measuring instruments, operating skills, and time cost. This paper proposed a fast identification method with non-integer exponents for PDGEs using double ball bar. Firstly, the error models in each plane are separately established. Next, the errors are pre-fitted by non-integer exponents with respect to the position of each linear axis, and the identification model is built based on that. Then, the length changing of the double ball bar is obtained by conventional circular tests in three orthogonal planes. Finally, 51 fitting coefficients of the pre-fitted errors are solved, by which the traditional direct solution of the specific value for PDGEs is replaced and the geometric error identification for the linear axis is realized. The series experiments have been carried out, and high prediction accuracy can be achieved with the deviation less than 2 μm and the root mean square error ( RMSE ) of 0.72 μm, which shows the validity of the proposed method. Graphicasl abstract: Image, graphical abstract