Artificial Neural Network-Based Quick and Robust Technique for Ottoman Minarets’ Fundamental Frequency Prediction


Nguyen Q. T., LİVAOĞLU R., Vu V. T.

3rd International Conference on Structural Health Monitoring and Engineering Structures, SHM and ES 2023, Da Nang city, Vietnam, 20 - 21 July 2023, vol.460 LNCE, pp.175-184 identifier

  • Publication Type: Conference Paper / Full Text
  • Volume: 460 LNCE
  • Doi Number: 10.1007/978-981-97-0399-9_17
  • City: Da Nang city
  • Country: Vietnam
  • Page Numbers: pp.175-184
  • Keywords: Artificial neural network, Frequency estimation, Masonry minarets, Ottoman minarets, Structural dynamics
  • Bursa Uludag University Affiliated: Yes

Abstract

Minarets are symbolic and indispensable parts of mosques in religious countries around the globe. Masonry minarets constructed during the Ottoman Empire are highly vulnerable to lateral excitations due to their brittle materials, slenderness, and distinctive shapes. After many minarets collapsed, preserving surviving ones in old Ottoman lands prone to earthquakes has become urgent. More importantly, determining the most vulnerable minarets to strengthen and maintain is a priority. Based on the earthquake spectrum of each region, the seismic vulnerabilities of minarets built on it can be warned effectively via the frequencies at some lowest modes. Instead of formulating the relationship between known geometrics and material information consisting of equivalent height, cross-section, inertia moment, Young’s elastic modulus, and mass density as considered in previous studies, this study presents an artificial neural network (ANN)-based approach to anticipate the fundamental frequency of Ottoman masonry minarets promptly. The measurements conducted on real minarets built around Bursa City (Türkiye) under ambient conditions are considered the output database of neural networks. Meanwhile, the available distinct parameters of minarets are selected to build the input dataset. As a result, the proposed ANN tool is practical and robust when the desired modal information can be estimated with high accuracy levels.