BROWSE

Related Researcher

Author's Photo

Shin, Myoungsu
Sustainable Structural Systems Lab
Research Interests
  • Seismic design, Tall buildings, High performance composites, Ultrasonic non-destructive evaluation technologies

ITEM VIEW & DOWNLOAD

Prediction of seismic drift responses of planar steel moment frames using artificial neural network and extreme gradient boosting

Cited 0 times inthomson ciCited 0 times inthomson ci
Title
Prediction of seismic drift responses of planar steel moment frames using artificial neural network and extreme gradient boosting
Author
Nguyen, Hoang D.Dao, Nhan D.Shin, Myoungsu
Issue Date
2021-09
Publisher
Pergamon Press Ltd.
Citation
ENGINEERING STRUCTURES, v.242, pp.112518
Abstract
This study aims to develop machine learning (ML) models that can predict the seismic responses of planar steel moment-resisting frames subjected to ground motions. For this purpose, two of the most powerful ML techniques, artificial neural network (ANN) and extreme gradient boosting (XGBoost), were applied. To generate a comprehensive dataset for the training and testing of the ML models, 22,464 nonlinear dynamic analyses were conducted on 36 steel moment frames with different structural characteristics (i.e., number of stories, number of bays, column-to-beam moment capacity ratio) subjected to 624 ground motions. The selected ground motions had peak accelerations greater than 0.2 g and earthquake magnitudes of at least 4.0. The maximum top and interstory drifts were considered as the primary seismic responses of the steel frames. The ANN and XGBoost models were developed using MATLAB and XGBoost 1.1.1 Package, respectively. The results suggest that both the ANN and XGBoost models could reliably estimate the seismic drift responses of the steel frames. The XGBoost model achieved a better prediction than the ANN model in all the considered cases. The coefficients of determination ( of the XGBoost model for the testing dataset are 0.975 and 0.962 in the maximum top and interstory drifts, respectively. The significance of the input variables on the prediction of the seismic drift responses was also analyzed. The ground motion intensities (e.g., peak ground acceleration, velocity, displacement) had a more significant impact on the seismic drift response prediction compared with the earthquake (e.g., magnitude) and soil characteristics, and the spectral accelerations in the first three natural periods. In particular, the peak ground velocity had the most significant effect on the seismic drift responses prediction. Finally, a graphical user interface based on the XGBoost model was developed for the preliminary estimation of the seismic drift responses of steel moment frames.
URI
https://scholarworks.unist.ac.kr/handle/201301/52895
URL
https://www.sciencedirect.com/science/article/pii/S0141029621006684?via%3Dihub
DOI
10.1016/j.engstruct.2021.112518
ISSN
0141-0296
Appears in Collections:
UEE_Journal Papers
Files in This Item:
There are no files associated with this item.

find_unist can give you direct access to the published full text of this article. (UNISTARs only)

Show full item record

qrcode

  • mendeley

    citeulike

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.

MENU