Deep Learning Enhanced Guided Wave Technique for Façade Debonding Evaluation

This thesis develops advanced techniques to detect hidden debonding in tiled façade systems using ultrasonic guided waves. It uses simulations and laboratory experiments to show how wave signals change when defects are present. The research introduces imaging and deep learning methods that can locate and quantify damage without dismantling the structure. It also designs a temperature compensation strategy to ensure reliable inspection under varying environmental conditions. The outcomes provide a more accurate and practical approach for assessing façade safety and reducing the risk of tile detachment in real buildings.