Abstract
This study presents an entropy-based framework for segmenting and characterizing the deformation behavior of AZ31B magnesium alloy sheets using acoustic emission (AE) signals obtained during tensile and V-bending tests. Shannon entropy was calculated from four AE parameters (duration, count, signal energy, and frequency centroid). The entropy values were normalized and averaged to identify transitions between deformation stages. Individual AE waveforms were transformed into normalized time–frequency images using the continuous wavelet transform (CWT), and quantitative features extracted from these images were used for Random Forest classification of the entropy-defined stages. Internal five-fold cross-validation yielded accuracies of 94.6% and 96.42% for tensile and V-bending deformation, respectively, indicating that the stages were distinguishable in the CWT-based feature space. Hierarchical density-based spatial clustering of applications with noise (HDBSCAN) separated fracture-associated AE signals during tensile deformation. The results indicate that the proposed method can characterize deformation-stage evolution and provide additional information on fracture-associated AE responses in AZ31B sheets.