Magnetostrictive Sensor Based Bridge Health Assessment Using Machine Learning Classifiers

Bridge Health Monitoring (BHM) is a crucial research domain in engineering, involving the analysis of infrastructure conditions like bridges using sensor data. Structural Health Monitoring (SHM) systems comprise distinct phases, with pivotal stages being feature extraction and pattern recognition. Signal processing techniques aid in feature extraction, while Machine Learning algorithms play a crucial role in pattern recognition, all geared towards accurately assessing bridge health. Applying Machine Learning techniques permits the automation of the SHM procedure and intelligent damage detection. This study introduces an innovative method utilizing Machine Learning technologies, leveraging data from Galfenol-based Magnetostrictive Sensors. These sensors convert vibrations into electrical data, offering insights into bridge health. A Decision Tree classifier categorizes data with an 80:20 split for training and testing, achieving a 98.72% accuracy. The code also demonstrates its predictive capability with new experimental data. Furthermore, various classifiers, including the Support Vector Method (SVM) and Neural Network, are tested, revealing their comparable accuracy. This research provides a user-friendly avenue for SHM and assists researchers in selecting optimal classification methods.