Fatigue damage assessment using adaptive acoustic emission waveform analysis

Abstract

Approximately 90% of mechanical failures stem from fatigue, and acoustic emission (AE) monitoring has shown promise in evaluating such damage. AE signal characteristics, such as event count, amplitude, and hit rate, are directly linked to fatigue progression and allow real-time tracking of critical damage stages. However, traditional feature-based methods often suffer from noise and interference from work hardening and user-defined settings, reducing accuracy. This study introduces an adaptive threshold waveform processing method that filters noise and enhances high-energy events. Analyzing these signals with the Bhattacharyya coefficient (BC) enables real-time fatigue assessment. Fatigue tensile-compression tests were conducted on medium-carbon steel, with concurrent real-time recording of surface temperature changes in the specimens. During the slow crack growth phase, an increase in local temperature corresponded with a turning point in the BC evolution trend, indicating that this method can reliably reflect the fatigue damage state of structures in real time. The effectiveness of the proposed method was validated through high-cycle fatigue experiments, demonstrating its applicability in practical fatigue damage scenarios. Furthermore, the computational cost analysis indicates that the proposed AHIE+BC framework achieves a data compression rate of approximately 20%, significantly reducing the computational burden while maintaining effective damage representation. These results highlight the method’s potential for efficient and accurate fatigue assessment in real-time structural health monitoring applications.