Akademik Veri Yönetim Sistemi
Structures, cilt.89, 2026 (SCI-Expanded, Scopus)
Historical masonry minarets are particularly vulnerable to dynamic events such as earthquakes due to their brittle construction materials and slender geometry. Regular Structural Health Monitoring (SHM) is essential for evaluating their condition and supporting conservation efforts. This study presents a hybrid Artificial Neural Network (ANN)-based methodology for damage assessment of masonry minarets. The ANN-based procedure comprises two primary stages: model calibration using a proposed iterative updating technique and damage localization. The proposed method is validated using a 3.625 m-high laboratory-constructed masonry minaret specimen, instrumented with three uniaxial accelerometers along its height, which capture variations in the vibration characteristics of the first two bending modes under ambient conditions. Results show that the calibrated numerical model achieves high accuracy in representing the real specimen, with discrepancies in the first two bending-mode frequencies remaining below 4% after calibration, and approximately only 2.6% for the fundamental mode, producing reliable databases for ANNs to enable effective early-stage damage detection. This hybrid ANN-based approach proves reliable and practical for earthquake engineering applications, leveraging limited modal data under ambient conditions to detect typical damage patterns of masonry minarets. By integrating artificial intelligence with physical modeling, the study highlights the potential for scalable solutions that support the preservation and resilience of historical masonry structures.