Ultrasonic-based defect identifcation in concrete using uniform manifold approximation and projection (UMAP)

Abstract

Damage identification using non-destructive techniques can aid in structural health monitoring and rehabilitation applications, especially in common construction materials like concrete. This paper attempts to provide a proof-of-concept, for the application of a novel feature extraction algorithm based on Uniform Manifold Approximation and Projection (UMAP), to detect highly common strength impacting defects such as cracking in concrete. To this aim, ultrasonic testing is adopted to transmit pressure waves through discretised locations along the specimen. The transmitted time signals are received by 54 kHz receivers and analysed with UMAP, resulting in highly reliable data separation between healthy and damaged sections–showing better results compared to competing wavelet decomposition frameworks. The key contribution of the study lies in the application of UMAP, for the first time, for damage detection in unreinforced concrete. This proof-of-concept demonstration with a positive outcome can lead to future investigations, which may delve deeper into the algorithm’s efficacy with various other defect types within concrete and reinforced concrete.