A new damage diagnosis approach for NC machine tools based on hybrid Stationary subspace analysis

12th International Conference on Damage Assessment of Structures

Gao, Chen^1 ; Zhou, Yuqing^1 ; Ren, Yan^1

College of Mechanical and Electrical Engineering, Wenzhou University, Wenzhou

325035, China^1

关键词: Damage Identification;    Empirical analysis;    Least squares support vector machines;    Phase space reconstruction;    Prior information;    Stationary components;    Statistic parameters;    Time and frequency domains;   

Others  :  https://iopscience.iop.org/article/10.1088/1742-6596/842/1/012047/pdf
DOI  :  10.1088/1742-6596/842/1/012047

来源: IOP

【 摘 要 】

This paper focused on the damage diagnosis for NC machine tools and put forward a damage diagnosis method based on hybrid Stationary subspace analysis (SSA), for improving the accuracy and visibility of damage identification. First, the observed single sensor signal was reconstructed to multi-dimensional signals by the phase space reconstruction technique, as the inputs of SSA. SSA method was introduced to separate the reconstructed data into stationary components and non-stationary components without the need for independency and prior information of the origin signals. Subsequently, the selected non-stationary components were analysed for training LS-SVM (Least Squares Support Vector Machine) classifier model, in which several statistic parameters in the time and frequency domains were exacted as the sample of LS-SVM. An empirical analysis in NC milling machine tools is developed, and the result shows high accuracy of the proposed approach.