Acoustic emission as a reliable technique for filiform corrosion monitoring on coated AA7075-T6: Tailored data processing

Abstract

Acoustic emission (AE) was used for in-situ filiform corrosion (FFC) monitoring on coated AA77075-T6. The analysis of AE data using DBSCAN as clustering algorithm (validated by Bhattacharyya Coefficients´ evaluation) has revealed the presence of three clusters (out of four) related to phenomena involved in the FFC mechanism: metal-coating interface delamination due to opening (tensile), sliding (shear) and mixed mode enclosing both previous ones. The peak frequency was found to be the most relevant descriptor for clustering by using Random Forest classifier, and the correlation with the dominant frequencies range was validated obtaining the Power Spectrum Density of the AE signals.