A MOVING SHIFTING VECTOR METHOD FOR RELIABILITY-BASED DESIGN OPTIMIZATION USING EFFECTIVENESS CHECKING OF PROBABILISTIC CONSTRAINT

Reliability-based design optimization (RBDO) aims at making the best decisions when unavoidable efforts of uncertainties are taken into account. To generate an accurate RBDO solution using as small number of function evaluations as possible, a moving shifting vector method is proposed in this work. Reliability index combined with a moving scheme are used to calculate the optimal most probable point and shifting vector, which is much simpler than the original optimal shifting vector method. Moreover, the result of reliability analysis in the proposed method is used to check the effectiveness of probabilistic constraint, which results in a novel constraint handling scheme with the aim of simplifying the deterministic optimization model in MSV. The new MSV method is very efficient for nonlinear problems. Two mathematical problems and a multi-component Tailor-welded blank structure crashworthiness design application are tested to show considerable savings of function evaluations in the proposed methods