This paper is concerned with a two steps computational approach for structural system reliability analysis: First, the search for points of maximum likelihood is reduced to an unconstrained optimization probem, using a conditional expectation techniqued coupled with an adaptive response surface interpolation function, defined over the integration domain and successively updated to take into account the previously calculated finite regions associated with local maxima. Second the failure probability is obtained by an effective utilization of the multimodal More Carlo importance sampling variance reduction technique over a finite number of discrete regions on the integration domain. Some results illustrate the main features of the proposed approach.
展开▼
