exit

Engineering and Systems   > Home   > Uncertainties and Reliability of Multiphysical Systems   > Optimization and Reliability   > Article

Overview of Structural Reliability Analysis Methods — Part III: Global Reliability Methods

Overview of Structural Reliability Analysis Methods — Part III : Global Reliability Methods


ChangWu Huang
INSA Rouen

Abdelkhalak El Hami
INSA Rouen

Bouchaïb Radi
LIMII FST
Settat - Morocco



Published on 9 February 2017   DOI : 10.21494/ISTE.OP.2017.0117

Abstract

Résumé

Keywords

Mots-clés

In Part III of the overview of structural reliability analysis methods, global reliability methods, which are based on global approximation model of performance function using Gaussian process model, are reviewed. Gaussian process model is the basis for these global reliability methods. This category of methods, firstly, approximates the performance function by Gaussian process model, and then perform sampling methods based on the built surrogate model to calculate the failure probability. The computational cost is significantly reduced with the aid of surrogate model, since the surrogate model is cheap to evaluate. Additionally, global reliability methods can give accurate results because Gaussian process model can adequately model the nonlinear limit state function. After the introduction of Gaussian process model, two global reliability methods, EGRA and AK-MCS are described and illustrated by an example.

In Part III of the overview of structural reliability analysis methods, global reliability methods, which are based on global approximation model of performance function using Gaussian process model, are reviewed. Gaussian process model is the basis for these global reliability methods. This category of methods, firstly, approximates the performance function by Gaussian process model, and then perform sampling methods based on the built surrogate model to calculate the failure probability. The computational cost is significantly reduced with the aid of surrogate model, since the surrogate model is cheap to evaluate. Additionally, global reliability methods can give accurate results because Gaussian process model can adequately model the nonlinear limit state function. After the introduction of Gaussian process model, two global reliability methods, EGRA and AK-MCS are described and illustrated by an example.

Reliability Analysis Gaussian Process Model Global Reliability Methods Monte Carlo simulation

Reliability Analysis Gaussian Process Model Global Reliability Methods Monte Carlo simulation