Application of clustering methods for assigning classes to acoustic signals

Evaluation of an acoustic emission waveforms acquired during structural health monitoring relies on the extraction and classification of features through the use of advanced digital signal processing. Trending of waveform features can permit the identification of damage types, propagation rate and severity. The analysis of trends within these extracted waveform features is often performed manually, possibly leading to inconsistencies in the information extracted. Accurate identification and assignment of classes is vital for the effective assessment of damage key characteristics of interest. The present study investigates the suitability of various clustering techniques with the aim of developing a fully automated system, which is independent of human input.