02/08/2022
By Noah Van Dam
Seminar Title: Automated damage detection and assessment in buildings and infrastructure using infrared thermography and deep learning techniques
Abstract:
The complexity of damage of structural components depends on the material as well as the construction process. Thus, when assessing or detecting damages using non-destructive techniques, it is crucial to consider this complexity in the analysis. Inspection and assessment of damages in large structures such as buildings and bridges, photogrammetry prove to be one of the widely applicable techniques. Since Ir images make the subsurface heat loss visible, adding these to photogrammetry improves the detectability of the subsurface damages. This tool is highly useful to assess the condition of a large-scale system by developing a virtual reality three-dimensional (3D) model. In this analysis, a 3D reconstruction approach is used to reconstruct a 3D model of various energy systems from Infrared (IR) images. The number of images and their overlap has some effect on the accuracy of the reconstructed model, especially when IR images are used. Understanding these effects is crucial for performing accurate structural health monitoring on the targeted systems. Another aspect is to process the 2D IR images which show the heat loss that can further be processed to detect the damage or irregularity in the structure. Artificial Intelligence is one of the ways of using this information to develop an effective diagnostics and prognostics framework. Thus, this study attempts to provide a solution for this research gap by providing a diagnostics framework based on deep neural networks.
Speaker Biography:
Shweta Dabetwar is currently working as a postdoctoral research associate at the University of Massachusetts, Lowell in the lab of Structural Dynamics & Acoustic Systems Laboratory with Alessandro Sabato. She has completed her doctorate at Texas Tech University in the department of mechanical engineering with Dr. Ekwaro-Osire. She is from India and has completed her baccalaureate studies there. Her research interests include structural health monitoring, machine learning, composite materials, artificial intelligence, photogrammetry, infrared imaging, and prognostics.