Automatized determination of the end of a wave mode in acoustic emission signals

Acoustic emission signals can be composed of multiple wave modes with different characteristics. The various wave modes contain information about their source, as it is related to the orientation of a source, the type of source and its depth in a plate-like structure. Furthermore, in structures with multiple possible wave modes, inversion of a wave can be impossible without mode recognition. Manual wave mode recognition is possible but tedious for a large number of signals. In previous work, the author tested a method for automating wave mode recognition. A shortcoming of this method is that it recognizes only the arrival of the wave modes. Subsequently, all samples from the start of a specific mode are attributed to this specific mode until the start of the next mode or until the end of the signal. This article presents an improved method for determining the end of a wave mode. As is already required for determining the starting moments of a wave mode, cross-correlations are calculated with a reference wavelet representing this wave mode. To determine the end of a mode, the next cross-correlation peak that follows the peak indicating the start of the mode is considered. This cross-correlation peak indicates either a repetition of the same wave mode, or the arrival of a high amplitude wave component belonging to a different wave mode. Hence, it is meaningful to consider such a cross correlation peak as an indication of the end of the wave mode in the signal.