A review on vision-based deep learning techniques for damage detection in bolted joints

Abstract

Bolted connections are widely used in steel structures. Detection of bolt loosening is the prime concern in the bolted joints to avoid sudden failure leading to catastrophe. Loosening of the bolts causes interfacial movement by reducing the pre-torque when subjected to vibrations due to dynamic loads. With the advent of computing capabilities, sensor technologies, and machine learning model accuracy in bolt loosening detection, damage recognition efficiency in bolted joints has increased. Integrating deep learning with machine vision, effective models can be proposed without human interventions. The present paper summarizes the research review on bolt loosening detection using machine vision and deep learning techniques from the past decade.