Advanced deep learning models for anomaly detection in masonry structures

4th International Civil Engineering & Architecture Conference, Trabzon, Türkiye, 17 - 19 Mayıs 2025, cilt.1, ss.358-368, (Tam Metin Bildiri)

The detection of structural anomalies is essential for safeguarding the integrity and longevity of

buildings with cultural and historical significance, such as masonry minarets. This study explores the use of

Convolutional Neural Networks (CNNs) for forecasting vibration responses and identifying patterns indicative of

potential structural damage. Vibration data collected from a minaret under both undamaged and controlled damage

scenarios are used to train and evaluate a CNN model designed to extract spatial features from time-series signals.

The model demonstrates strong performance in distinguishing normal structural behavior from anomalous

patterns, offering high accuracy and sensitivity in damage detection. In addition to its predictive capabilities, the

CNN approach proves to be computationally efficient and suitable for real-time monitoring applications. The

findings highlight the potential of deep learning–based methods, particularly CNNs, to enhance non-contact

structural health monitoring strategies, providing a reliable and scalable solution for the preservation of historic

structures.