Improving of Acoustic Emission Signal Detection for Fatigue Fracture Monitoring

Abstract Identifying growing cracks in running machinery and industrial facilities is challenging, particularly at the nucleation stage in “hard-to-reach” places and in harsh environment. The method of acoustic emission is a popular non-destructive means for inspecting and monitoring the behavior of loaded materials and active internal defects. One of the key problems which impedes a wider application of the AE technique is associated with detectability of low amplitude signals hidden on a background of laboratory or industrial noise. A recently proposed continuous threshold-less mode of AE data acquisition offers an advantage in analyzing AE random time series on different time scales, thus providing information otherwise inaccessible by a conventional threshold-based acquisition mode. The evaluation criteria concerning the activity of internal defects depend strongly on the AE detectability in the selected time interval. In the present work we compare performance of two signal detection algorithms: conventional amplitude threshold discrimination and innovative «phase picker» based on a wavelet transform. In observing the characteristic accumulation of AE signals over long periods of hundreds time and opposite in a single loading cycle can provide a difference at signal estimation.