Release Time:2019-03-12  Hits:

Indexed by: Journal Article

Date of Publication: 2018-02-01

Journal: KSCE JOURNAL OF CIVIL ENGINEERING

Included Journals: Scopus、EI、SCIE

Volume: 22

Issue: 2

Page Number: 708-717

ISSN: 1226-7988

Key Words: artificial neural network; damage prediction; steel frame; seismic fragility

Abstract: Steel frames are widely used in earthquake areas but the fragility analysis of the steel frames is a time consuming process. The fragility curves are generally obtained based on a large number of time history analyses. In order to better simulate steel frames, the material and geometric uncertainties are considered. The moderate earthquake records are selected as the earthquake samples. Then the seismic damages induced by these samples are obtained using Finite Element Analysis (FEA) program. An Artificial Neural Network (ANN) is trained by these seismic damage data. Then the trained ANN is used to predict the seismic damage of the steel frame. When the training data is enough, the method has high prediction precision. The fragility curves based on the ANN prediction are the same as that based on the FEA program.