Self‐adaptive real‐time clustering analysis and damage pattern recognition for steel tube confined reinforced concrete structures through acoustic emission signals

Abstract

This paper provides a novel and effective self‐adaptive real‐time clustering (SARTC) strategy for clustering real‐world large datasets real time, and a novel feature selection method (LS‐MI) was proposed to enhance the clustering efficiency. The effectiveness of the novel methods was validated by theoretical analysis and experimental verification on three benchmark datasets; the result shows that the novel methods achieved the superiorities of high efficiency, high accuracy, and adaptive convergence. And the novel methods were applied to damage pattern recognition for steel tube confined reinforced concrete columns through acoustic emission (AE) signals; the result shows that the proposed LS‐MI procedure can retain AE features with strong representativity but low redundancy, while the SARTC strategy can classify the real‐time AE signals into three clusters with clear bonds. The generalized AE clustering structure was discussed, and possible relation of the clusters to the damage types were explicated; these results create a foundation for establishment of general AE interpretation rules for damage mode identification in future works.