Analysis of the configuration characteristics of the chaotic attractor for machine tool dynamic error under the singular spectrum

Abstract

To explore the chaotic characteristics of the motion error of CNC machine tools, this paper proposes a chaotic attractor feature extraction method for the early motion error of CNC machine tools. Firstly, the chaotic verification of each time series feature of machine tool motion error proves that the change of CNC machine tool motion error is a nonlinear change process with chaotic characteristics. It then establishes the multivariate error phase space and performs a singular spectrum analysis. Because the motion error is strongly disturbed by noise, it is difficult to determine the number of singular values by the traditional singular spectrum analysis method, and this paper proposes to calculate the phase space index of the reconstructed sequence by traversing the number of singular values, to determine the main feature components; Finally, under different noise environments and different signal lengths, three feature extraction methods are introduced for experimental comparison with this paper's method, which shows that the chaotic attractor reconstructed after determining the number of singular values has a more apparent geometric structure. The method proposed in this paper has advantages in machine tool motion error.