Machine Learning Clustering techniques applied on Acoustic Emission data: an exploratory approach

Acoustic Emissions (AE) techniques are a non-destructive testing aiming at monitoring and researching the health of many materials, especially construction ones, like prestressed reinforced concrete elements. In order to evaluate material properties, stability, and strength over time, in fact, an early identification of micro-fractures by depicting the release of mechanical energy of an elastic wave radially propagating through the material from the fracture zone can reveal in advance potentially dangerous damage. Several classification approaches can be found in the literature for such a type of acoustic data, but classification approaches based on Machine Learning techniques are neglected, especially for AE applied to reinforced concrete. In this contribution, a preliminary analysis on clustering of Acoustic Emission data measured during a four-point bending test on a prestressed reinforced concrete beam is presented. Results indicate a possible innovative way for the reading of the outputs of Acoustic Emissions.