A Novel Rust Damage Diagnosis Method for Reinforced Concrete Based on WAESN and GADF-AlexNet-Multihead-Attention

Upon corrosion of the rebar within concrete, there is a substantial reduction in the effective cross section of the rebar, leading directly to decreased strength of the rebar itself, cracking of the concrete, weakened bond strength between the rebar and concrete, and spalling of the concrete cover layer. These issues pose a significant threat to public safety and property. This article presents the design and implementation of a reinforced concrete corrosion diagnosis system based on a wireless acoustic emission sensor network (WAESN), primarily aimed at nondestructive testing of rebar corrosion. Initially, WAESN nodes perform high-speed sampling of the acoustic emission (AE) signals from rebar concrete specimens under electrified accelerated corrosion, calculate characteristic parameters, and transmit them wirelessly to a gateway. Subsequently, the Gramian angular difference field (GADF) is utilized to transform 1-D time series data into polar coordinates, effectively capturing the nonstationary time-variant characteristics and the differences between modes, thus maximizing the distinction among various features and profiling the deeper aspects of concrete rebar corrosion. Finally, by integrating Multihead-Attention, the classification accuracy of AlexNet is enhanced, allowing for precise detection of different corrosion stages. The proposed GADF-AlexNet-Multihead-Attention model achieved accuracies, precision, recall, and <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">${F}1$ </tex-math></inline-formula>-scores of 97.42%, 97.68%, 98.13%, and 97.55%, respectively, surpassing other comparative algorithms and demonstrating superior monitoring capabilities.