Inverse Reliability-based Optimization

A new inverse reliability-based optimization method is presented. It is aimed for solving inverse reliability problems with multiple solutions of optimal design parameters with respect to the target reliability constraints. The method is based on coupling of aimed multilevel sampling optimization method and artificial neural network-based inverse reliability method. Emphasis was placed on the reduction in computational effort, which is further enhanced by using a small-sample simulation technique. A unique optimum solution is obtained by introducing an optimization with an objective function related to deterministic constraint while reliability constraints are implicitly considered via inverse reliability analysis. The validity and efficiency of the method is shown on testing example with three design parameters and one reliability constraint. Both cases with uncorrelated as well as correlated random variables have been tested.