In this paper a novel approach is proposed to solve the reliability based design optimization (RBDO) problem, which is translated into a single-level problem using Karush-Kuhn-Tucker (KKT) conditions. The transformation of a bi-level RBDO problem into a single-level problem using KKT conditions introduces several equality constraints in the single-level problem definition. Presence of multiple equality constraints poses numerical difficulty to the gradient based optimizers, hence a robust algorithm to solve the single-level RBDO problem is proposed in this paper using an alternative approach. The proposed approach uses an exterior penalty based cross-entropy (CE) method to solve the uni-level RBDO problem. This approach is shown to be robust in handling equality constraints. The three example problems solved in this paper also shows that the algorithm works well with different starting points used for the design variables.

This content is only available via PDF.
You do not currently have access to this content.