Acoustic emission localization in thin multi-layer plates using first-arrival determination

In the case of thin plates, there exist two modes of propagation which travel at different velocities and exhibit dispersion characteristics. Techniques that are based on Gabor wavelet transform or cross-correlation technique are commonly used to locate acoustic emission (AE) events which occur in large plates. Due to side-edge reflections and short source-to-sensor distances, these techniques are not suitable for small plate-like specimens. If the thickness of the plate-like specimen is smaller than a specific value, the first-coming (extensional) mode will show non-dispersive behavior in AE frequency range. Under such a condition, the conventional localization method can be used for detecting first-arrival times on non-dispersive extensional mode. In previous paper, authors of the paper developed a first-arrival automatic determination technique based on Akaike information criterion (AIC) for thin metal plates. This paper compares this technique with another AIC approach, STA/LTA method (short-term average/long-term average) and a standard threshold-crossing technique. The comparative analysis includes blind tests, and is provided on four datasets recorded by a four-channel recording system. The three of four datasets were generated using two types of artificial AE sources (Hsu–Nielson source and laser impulse), while the fourth one contains real-measurement data. Each dataset corresponds to measurement made on a thin-plate specimen of a different material or geometry.

► First-arrival determinations of extensional modes could estimate source locations.
► The choice of characteristic function is a crucial factor for the first-arrival detection.
► Parameters of approaches were optimized by a dataset and tested on other datasets.
► Two-step AIC approach showed the best performance in comparison to other techniques.