Prototype validation of a deep learning-based bridge health monitoring framework using Galfenol Magnetostrictive sensors

Abstract

Ensuring the structural integrity of bridges is critical for safety, operational efficiency, and lifecycle sustainability of infrastructure. Traditional inspection techniques are often manual, labour-intensive, and reactive, lacking the precision needed for proactive maintenance. This study presents an automated Structural Health Monitoring (SHM) framework tailored for bridge infrastructures, combining Galfenol-based Magnetostrictive Sensors (MsS) with state-of-the-art Deep Learning (DL) algorithms. The system enables real-time, self-powered monitoring by capturing vibration-induced electrical signals and classifying structural conditions autonomously. Six machine learning models, namely Random Forest, Deep Neural Network (DNN), Recurrent Neural Network (RNN), Gated Recurrent Unit (GRU), Long Short-Term Memory (LSTM), and Convolutional Neural Network (CNN), were evaluated on data from a scaled beam bridge prototype. Among them, the LSTM achieved a peak classification accuracy of 99.5%. The proposed system contributes to automation in construction by enabling intelligent condition assessment, reducing dependence on manual inspections, and supporting predictive maintenance strategies. This integration of smart sensors and AI facilitates continuous, non-invasive infrastructure monitoring, aligning with the goals of automated facility management and sustainable construction practices.