Machine Learning-Based Corrosion-Like Defect Estimation With Shear-Horizontal Guided Waves Improved by Mode Separation

Shear Horizontal (SH) guided waves have been extensively used to estimate and detect defects in structures like plates pipes. Depending on the frequency plate thickness, more than one guided-wave mode propagates, which renders signal interpretation complicated due mixing complex behavior of each individual interacting with defects. This paper investigates use machine learning models analyse two lowest order SH modes, for quantitative size estimation detection corrosion-like aluminium plates. The main contribution present work is show that separation through improves effectiveness predictive models. Numerical simulations performed generate time series creating estimators, while experimental data validate them. We a full scheme decreased error rate final model by 30% 67% defect respectively.