Neural Network-Based Estimation of Seismic Demand Parameters for Reinforced Concrete Frame Buildings

Date

2025-4-25

Author

Kaya, Yunus Eren

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

331
views

0
downloads

Cite This

Evaluating structures’ seismic performance is crucial for ensuring life and property safety in earthquake-prone regions. Recent earthquakes have shown that design and construction errors can significantly increase earthquake losses. Traditional assessment methods range from rapid, preliminary techniques to detailed nonlinear analyses. However, simplified methods often produce overly conservative results due to limited data and idealized assumptions, while detailed methods require extensive data and high computational costs. This thesis aims to present a sufficiently accurate framework for large-scale seismic assessment applications. To this end, a multi‑level neural network-based approach is proposed to estimate key seismic demand parameters of reinforced concrete (RC) frame buildings. Three distinct neural network models with differing input parameters are trained to estimate the fundamental structural period and maximum interstory drift ratio (MIDR). For training, a comprehensive synthetic dataset of 795,060 buildings is generated from an inventory of 4,417 RC frame buildings by systematically varying seismicity levels, concrete‑strength values, and soil classes. Explainable Artificial Intelligence techniques are employed to interpret model predictions. The validity of the neural network is demonstrated by comparing demand categories based on predicted MIDR values with observed damage data from the 2023 Türkiye earthquakes. Finally, the model is applied to a simulation dataset of 23 million buildings, producing a nationwide seismic demand map based on predicted MIDR values. The developed approach significantly outperforms traditional methods in terms of speed and accuracy, providing a practical tool for rapid vulnerability assessment and risk-mitigation planning for RC frame buildings.

Subject Keywords

Data-Driven Seismic Demand Prediction, Artificial Neural Networks, Maximum Interstory Drift Ratio, Reinforced Concrete Buildings, Explainable Artificial Intelligence

URI

https://hdl.handle.net/11511/114954

Collections

Graduate School of Natural and Applied Sciences, Thesis