Artificial Neural Networks for seismic demand prediction of a single degree of freedom
Abstract views: 60 / PDF downloads: 89
DOI:
https://doi.org/10.59287/icaens.995Keywords:
Seismic Demand, Artificial Neural Network, Incremental Dynamic Analysis, Nonlinear Time History AnalysisAbstract
This paper proposes using an artificial neural network (ANN) to estimate and predict the seismic demand of Single Degree of Freedom (SDOF) systems. Our methodology entails the production of a comprehensive dataset containing SDOF and earthquake characteristics. Nonlinear Time History Analysis (NL-THA) is performed on a randomly generated SDOF system using thirty-one artificial ground motions (GMs) matched to the EuroCode-8 (EC8) response spectrum to train the ANN model. To assess the performance of the ANN model, we compare the Incremental Dynamic Analysis (IDA) curves, the median IDA curve, and the 3D fragility surface in a case study. This analysis assists in determining the precision and dependability of the predicted maximum displacement of the SDOF system. The results showed a remarkable reduction in processing time without losing prediction accuracy. . It was concluded that the ANN-based method can be used as an alternative for the current method for estimating the performance points and the fragility assessment of buildings.