A novel machine learning based approach for health monitoring of concrete structures

Abstract

Introduction: The health of the concrete structures was initially inspected through visual inspection or using tools to verify the damage and health of the concrete structures. The machine learning models altered this process where the images of the concrete structures are used as inputs and passed through the model to identify the cracks and predict the level of damage to prevent accidents and further damages. The cracks are formed in the concrete structures because of the constant expansion and contraction irregularities. These irregularities are deemed as the potential damagesand#8223; that are caused naturally within the concrete structures in buildings, pavements, bridges, and other structures. The damages and irregularities found in the concrete structures are evaluated by civil engineers or inspection staffs manually which also increases time, incurs higher cost and budget and finally results in error where certain cracks are non-detectable. The usage of machine learning (ML) models and prediction approaches (image recognition, pattern recognition, image classification, image segmentation, and more) in the crack detection process are found as easier, rapid and robust. newline