Damage Pattern Recognition and Feature Extraction of MgO-C Refractory

摘要

The determination of the damage mode and the quantitative description of the damage of the refractory are the subjects of this paper. Through the three-point bending tests, a magnesia-carbon refractory has been investigated using an Acoustic Emission (AE) unsupervised pattern recognition procedure and a frequency-energy coupled analysis. By means of coefficient analysis of the wavelet energy spectrum, the damage is divided into three periods. Based on the periods, time-frequency analysis is used to extract the damage characteristics and the centroid energy carried by the centroid frequency is put forward in the explanation of the extent of the damage. Finally, the time history plot of the centroid frequency is produced. It shows the similarity of that of the energy which indicates that the extent of the damage could be explained by the centroid energy carried by the centroid frequency brought forward.