Loading...

Seismic Damage Estimation Considering Effective Structural Features through Nonmodel Approach and Machine Learning Techniques

Jamdar, Mahshad | 2024

0 Viewed
  1. Type of Document: Ph.D. Dissertation
  2. Language: Farsi
  3. Document No: 57785 (09)
  4. University: Sharif University of Technology
  5. Department: Civil Engineering
  6. Advisor(s): Mohtasham Dolatshahi, Kiarash
  7. Abstract:
  8. This research presents advanced methodologies for the seismic damage assessment of structures, focusing on the rapid and precise estimation of key engineering demand parameters. By integrating modern signal processing techniques with machine learning models, the study introduces a novel, data-driven framework that eliminates the reliance on traditional, computationally intensive modeling approaches. Leveraging real-time data from seismic sensors, these methodologies accurately predict critical performance metrics including residual and peak story drift ratios, peak floor accelerations, and fragility curves, significantly improving the accuracy of seismic assessments. The inclusion of real-world damage data enhances the robustness and precision of fragility estimations. Additionally, this study investigates the significant role of soil-structure interaction (SSI) in influencing the dynamic response of structures, offering a more comprehensive understanding of seismic behavior. To support the data-intensive nature of the proposed approach, an extensive database of structural responses to earthquake loading was developed through numerical simulations using OpenSees. This database covers a range of structural systems, including eccentrically braced frames (EBFs) and steel moment-resisting frames (MRFs), which were used to train machine learning and deep learning models, further improving the accuracy of damage prediction and performance evaluation. The proposed framework combines advanced signal processing with machine learning and deep learning techniques, substantially reducing the time required for seismic damage assessments while simultaneously increasing prediction accuracy. Furthermore, the use of wavelet-based damage-sensitive features using system identification enhances the precision of structural damage identification, positioning these data-driven methodologies as a compelling alternative to conventional seismic engineering models
  9. Keywords:
  10. System Identification ; Soil-Structure Interaction ; Machine Learning ; Fragility Curve ; Engineering Demand Parameters (EDP) ; Deep Learning ; Wavelet-Based Damage-Sensitive Features ; Fast Detection ; Vulnerability Estimation ; Seismic Assessment

 Digital Object List

 Bookmark

No TOC