Volumetric error modeling and accuracy improvement by parameter identification of a compound machine tool 1. IntroductionLarge-scale products such as aviation, automobile and mold have the structure features of complex curved surface, and the demand for five-axis machining centers of high-precision is increasing. Most of these products are thin-walled structures with large dimensional changes and require high precision for docking and assembly. The modeling and compensation of geometric errors are important ways of improving the accuracy of the machining center.The geometric error modeling of CNC machine tools is mostly based on rigid body kinematics combined with the homogeneous transformation matrix method, and the single geometric errors of the coordinate axis is introduced to establish the geometric error model of the machine tool. Ordinary three-axis machine tools must test 21 geometric errors (Okafor and Ertekin, 2000). The number of test error parameters of a typical five-axis machine tool with dual rotary axes is 52 (Schwenke et al., 2008). Fan et al. (2014 used the orthogonal polynomial regression method to obtain the spatial geometric error distribution and kinematic model of a five-axis machine tool. Chebyshev polynomials were also applied to fit position-dependent geometric errors, it contributed to the establishment of the integrated model and parameter error identification in three-axis machine tools (Li et al., 2015;Aguado et al., 2012).Recently, related research used screw theory to model and compensate for the geometric errors of machine tools. Xiang et al. (2016) proposed the forward and reverse kinematic model of a five-axis machine tool based on screw theory