Optimized Monte Carlo simulations for system reliability analysis

Within the reliability, availability, maintainability and safety (RAMS) analysis, it is of great interest the study of highly reliable systems. It is well-known that, through the Monte Carlo (MC) method, the study of such systems has to be tackled by a variance reduction technique [1]; here the so-called importance sampling (IS) has been considered. As far as the IS technique is concerned, we point out that the issue of optimization is still challenging and it is typically left to the MC user experience [2], hence one has usually to make many attempts before achieving an advantageous biasing. The present paper focuses on the unreliability estimate of systems in which failure events occur independently and shows a criterion allowing to get the set of optimizing parameters.