服役构件疲劳损伤的声发射信号特征提取

Feature Extraction of Acoustic Emission Signals for Fatigue Damage of In-service Components

基于拉-拉疲劳试验,在线监测了试件关键部位的声发射信号在不同疲劳循环次数下的变化。通过对声发射监测信号进行统计分析、小波包能量谱分析和小波熵特征提取,确定了反映疲劳损伤的声发射特征参数为幅值、电压、高频能量占比和小波熵值,结果表明:所提取的特征参数均将整个疲劳过程划分为初始、中间和后期3个阶段,较好地反映了疲劳寿命循环的裂纹萌生、裂纹稳态扩展和裂纹失稳扩展3个阶段,可用于不同疲劳寿命区间的预测。

Based on the tensile-tension fatigue test,the acoustic emission signals from key parts of the specimen under different fatigue cycles were monitored online.Through statistical analysis of acoustic emission monitoring signals and the wavelet packet energy spectrum analysis and wavelet entropy feature extraction,the characteristic parameters of acoustic emission like the amplitude,voltage,high frequency energy proportion and wavelet entropy which reflecting the fatigue damage were determined.The results showed that,the whole fatigue process can be divided into early stage,middle stage and late stage by the extracted feature parameters,and the proposed approach can be used to predict different fatigue life intervals because it better reflects three stages of fatigue life cycle including crack initiation,steady propagation and unstable propagation.