Validating a Physics-Based Automatic Classification Scheme for Impact Echo Signals on Data Using a Concrete Slab with Known Defects

Impact echo (IE) is capable of locating subsurface defects in concrete slabs from the vibrational response slab to a mechanical impact. For an intact (“good” condition), frequency spectrum IE dominated by single peak corresponding slab’s “thickness resonance frequency,” whereas presence (“fair” or “poor” conditions) could manifest various ways such as multiple distinct peaks at frequencies higher, lower, than thickness resonance. In previous research, authors have proposed partitioning for signal classification. Firstly, band identified using data-driven approach and then signals are represented their energy distribution three bands—frequencies less than, within, greater Following this feature extraction, unsupervised clustering used identify centroids each class—good, fair, poor—which further classify any test into one aforementioned classes. The classification developed training on unlabeled real bridge deck data (the Federal Highway Administration’s [FHWA’s] InfoBridge dataset) without making use labeled data. This study aims validate methodology dataset eight reinforced specimens constructed FHWA Advanced Sensing Technology Nondestructive Evaluation laboratory having known artificial defects. Our findings indicate that physics-based definition method robust can with moderate accuracy.