Separation method for acoustic emission signals of prestressing steel strand fracture and concrete cracking

In the monitoring of strand fracture in prestressed concrete bridges, concrete cracking often becomes severe, interfering with the identification of strand fracture signals and limiting the effectiveness of monitoring. To address this issue, this paper proposes a novel acoustic emission (AE) signal separation method. By extracting and analyzing the features of strand fracture signals and concrete crack signals, and utilizing K-means clustering combined with support vector machines (SVM), efficient separation of the two signal types is achieved. The results indicate that peak frequency, amplitude, and duration are the key parameters for distinguishing these signals, with the combination of duration and amplitude optimizing classification. The proposed method achieves over 99% accuracy in signal recognition, offering effective technical support for the health monitoring of critical structures.