Implementing Artificial Neural Networks and Mode Shape Curvature for Locating damage in Bridge Structures

Abstract

Structural damage manifests as changes in a system’s geometry and material properties, leading to stiffness reduction and adversely impacting performance. These reductions alter modal parameters like natural frequencies and mode shapes, which can be analysed to identify damage. Modal analysis allows for the extraction of modal frequencies and shapes, enabling a detailed examination of mode shape curvature to locate structural issues accurately. Recently, artificial neural networks (ANNs) have proven to be highly effective in structural health monitoring, particularly due to their exceptional pattern recognition capabilities. This study presents a novel approach that combines mode shape curvature analysis and ANNs to detect damage in steel girder bridge structures. Through vibration-based fault detection, this approach overcomes limitations of traditional methods by using mode shape curvature as a reliable indicator of structural anomalies. Experimental evaluations compare the modal responses of intact and damaged structures, providing critical insights into changes in structural behaviour. A feed-forward neural network with two hidden layers is employed, trained on damage indices generated from mode shape curvature data. This trained ANN is then used to identify unknown damage locations within the structure, demonstrating high precision. Validation results further confirm the accuracy and reliability of this damage detection method. Overall, the study shows that ANNs trained with modal curvature data offer a robust and efficient approach for early damage detection in steel girder bridge structures, significantly enhancing safety and operational reliability. This innovative method contributes to advancing structural health monitoring, offering a valuable tool for maintaining the integrity of infrastructure, such as steel girder bridges.