286 Diffuse Response Surface Model Based on Advancing LHS Patterns

Deterministic design optimization without taking uncertainties into consideration may lead to unreliable results. Reliability-Based Design Optimization (RBDO) is getting a lot of attention recently. However, RBDO is computationally expensive. We focus here on a response surface methodology adapted to the RBDO. The Diffuse Approximation (DA) based on a progressive sampling pattern is used within a variant of the First Order Reliability Method (FORM) for calculating the reliability index. An advancing Latin Hypercube Sampling (LHS) is proposed in this paper to maximally reuse previously computed points while adding a minimal number of new neighboring points at each step. Finally, we investigate two mathematic cases. The results show how the method proposed can reduce greatly the number of function evaluations and improve the computational efficiency.