Computation of Seismic Fragility Curves Using Artificial Neural Network Metamodels
Résumé
In earthquake engineering, the fragility curve is defined as the conditional probability of failure of a structure, or its critical components, at given values of seismic intensity measures (IMs). For simulation-based fragility curve estimations, this conditional probability of failure is usually computed with log-normal assumption. The artificial neural network (ANN) is used to improve the computational efficiency of the simulation-based fragility analysis. An ANN metamodel is built from 100 finite element soil-structure-interaction simulation results. The most relevant IMs are selected based on semi-partial correlation coefficients. The ANN metamodel is trained with the selected IMs, and a large number of Monte Carlo simulations are performed with this metamodel. Fragility curves are computed with both parametric (log-normal model) and non-parametric methods for the estimation of the risk of failure of an electrical cabinet in a reactor building studied in the framework of the KARISMA benchmark.
Origine : Fichiers produits par l'(les) auteur(s)
Loading...