Structural reliability assessment using Kriging metamodel and Monte Carlo simulation: AK-MCS method

摘要

This article describes a new method to perform structural reliability analysis. It is called AK-MCS for Active learning method combining Kriging and Monte Carlo Simulation. It proposes to perform a Monte Carlo simulation without evaluating the whole population on the expensive performance function. This is possible thanks to the predictions by Kriging metamodel which is based on a design of experiments of weak size. The method is performed on several examples and compared to other metamodels. It is proved to require less calls to the performance function and for a better accuracy on the probability of failure.