Keywords: reliability-based design optimization, adaptive sampling, meta-modeling, multi-objective optimization, non-dominated sorting genetic algorithm II, asymptotic sampling.

The goal of this paper is to formulate a methodology for the multi-objective reliability-based design optimization (RBDO) that is applicable for any type of simulation model. The double-loop RBDO formulation seems to be the most robust approach for any kind of a nonlinear limit state function, in which the reliability assessment is nested in the inner cycle corresponding to the optimized variables designed in the outer cycle. The applied asymptotic sampling as an adaptive sampling method for reliability assessment seems to be a good compromise between the costs and accuracy. The computational demands can be decreased by using a meta-model instead of the simulation model. Since the design variables change with every iteration and the meta-model is utilized for the reliability assessment, the meta-model is trained only in the vicinity of the relevant design variable which makes the meta-model computationally faster and more precise. Samples from the asymptotic sampling located close to the limit state dividing the space in the failure and the safe region along with a space-filling property are our candidates for the effective multi-objective update of the design of experiment which is used to form the meta-model.

purchase the full-text of this paper (price £20)

go to the previous paper
go to the next paper
return to the table of contents
return to the book description
purchase this book (price £50 +P&P)