Acoustic emission identification of wheel wear states in engineering ceramic grinding based on parameter-adaptive VMD

Abstract

In engineering ceramic grinding, the wheel wear states have an important influence on the processing efficiency and grinding quality. To address the difficulty of online monitoring of wheel wear states, a signal reconstruction method based on parameter-adaptive variational modal decomposition was proposed. Experiments on the entire life cycle of the grinding wheel were performed, and the wear states were classified based on the observed surface morphologies of the wheel and workpiece. Subsequently, acoustic emission signals were collected by nodes, and high correlation frequency bands were chosen and reconstructed. The signal features were extracted, and the wear states were identified based on a random forest optimization feature combination, finally. In this study, taking alumina ceramic as an example, eight groups of single-factor experiments were designed based on cumulative grinding depth to determine the optimal linear speed of the grinding wheel, worktable moving rate, and other process parameters. A total of 160 groups of AE signal samples were obtained. The results show that feature extraction by frequency bands can better reflect the changes in the grinding wheel wear states than the overall signal. In addition, optimization feature combinations based on the bands after decomposition, filtering, and reconstruction can divide and identify the various wear states more effectively. The comprehensive identification accuracy reached 90.6%. The accuracy of identifying the severe wear state reached 100%. Thus, the proposed method provides theoretical guidance for the practical industrial applications of online monitoring of wheel wear states.