Time-dependent reliability analysis framework combining active learning and ensemble of surrogates

We present some advances in the application of ensemble of surrogate models to the study of time-dependent reliability problems.We introduce a framework combining limit state functions with an appropriate temporal discretization and the training of instantaneous surrogates mapping the limit state function.This framework is an extension of the traditional active learning strategy widely used in classical (time-invariant) reliability.Our main result is the implementation of ensembles for this type of problem, and we compare our results with Monte-Carlo simulations and with a single-surrogate strategy based on the same active learning, time-discretization principle.