Non-Destructive Techniques for Bridge Inspection

Abstract

This research proposes a novel Visual Inspection (VI) technique for bridges referred to as Semi-Automated Virtual Reality (SAVR) Inspection via state-of-the-art 3D scanning, Artificial Intelligence (AI), and Virtual Reality (VR). Recent high-profile collapses coupled with an aging bridge stock, increased loading, and the pressures of climate change have led policymakers to focus on bridge management. VI is standard practice around the world to improve public safety. Only a few studies have attempted to address the limitations of VI. The SAVR Inspection technique aims to address all the limitations of the conventional VI approach using a five-step methodology. The first stage involves (i) creation of digital twins of the built environment via state of the art 3D scanning; (ii) development of VR inspection to enable 3I’s, immersive, interactive, and imaginative inspection experience from the comfort of the office; (iii) reduction of point cloud data using Weighted Self Organising Maps (wSOM), AI method, to increase the scale of the structure that can be inspected in the VR; (iv) the use of Convolution of Neural Networks, modified VGG-19 - AI method, for automated crack classification of reinforced concrete bridges per the Severity Extent Criteria, Design Manual for Road and Bridges (DMRB), UK; and (v) combining prior stages (ii), (iii) and (iv). This thesis explains the background, context, methodology, and results of each of these steps and discusses how each of these steps addresses the shortcomings of the conventional VI technique. It is concluded that the SAVR Inspection technique can greatly assist engineers in the inspection process by increasing the accuracy and addressing major limitations of the inspection regime. The work is an important step towards the automation of VI of bridges necessary for the safe and sound functionality of the infrastructure. The work will be of particular interest to academics and students, bridge engineers, construction professionals, law makers and could lead to future revisions of the bridge inspection process and standards.

Date of Award	31 Dec 2023
Original language	English
Awarding Institution	The University of Manchester
Supervisor	Mojgan Hadi Mosleh (Supervisor) & Lee Margetts (Supervisor)