A new health estimation model for CNC machine tool based on infinite irrelevance and belief rule base

Abstract

To guarantee the normal workflow and determine scheme of optimal maintenance, it is important to accurately estimate the health condition of computerized numerical control (CNC) machine tool. In current studies, the health condition of CNC machine tool is modeled by using one feature. Due to the complexity of CNC machine tool, the estimating accuracy of the current models is poor and real-time performance cannot be satisfied when multiple features are chosen. Moreover, it is difficult to obtain more effective monitoring data when the CNC machine tool is from normal to failure. To solve the problems, based on infinite irrelevance and belief rule base (BRB), a health estimation model which is named as the infinite irrelevance BRB model is proposed in this paper. In particular, the infinite irrelevance method is used to select key features to optimize the model structure, and BRB is applied to estimate the health condition according to the monitoring data and expert knowledge. Thus, the quantitative monitoring data and expert knowledge can be used effectively to improve accuracy and real-time performance of health estimation. Furthermore, because the initial values of the parameters in the proposed infinite irrelevance BRB model given by experts may not be accurate, the constraint covariance matrix adaptation evolution strategy (CMA-ES) algorithm is employed to train the parameters. A case study for the servo system of the CNC milling machine is used to verify the effective and accuracy of the proposed model. The results show that the infinite irrelevance BRB model can accurately estimate the health condition of the servo system.