Analysis and Clustering of Acoustic Emission Signals in the Tensile Deformation of AZ31B

The application of acoustic emission (AE) has applied to detect the yield and fracture of materials. In this study, the deformation characteristics of the magnesium alloy (AZ31B-H24) were characterized during tensile testing using AE signals. First, the AE signals of AZ31B-H24 sheets with thicknesses of 1 and 3 mm were investigated during tensile deformation. Numerous AE signals were generated during yielding and fracture, and their signal characteristics were analyzed. The signals for yield deformation and fracture deformation were observed to differ. The duration of the yield signal was longer than that of the fracture signal, and the energy of the yield signal was lower than that of the fracture signal. Based on these characteristics, the AE signals were categorized using the clustering method, an unsupervised learning algorithm, into four categories: Cluster 1 comprises the AE data obtained at the yield point of the magnesium alloy plate. Clusters 2 and 3 comprise those obtained in the stages from work hardening to failure. Finally, Cluster 4 comprises those obtained during the fracture point. The average value of each AE parameter was obtained. In the frequency domain, the peak frequency of the yield signal was higher than that of the fracture signal. The energy and amplitude of the signal were the highest in the fracture.