Sequential approximate robust design optimization using radial basis function network

摘要: This paper proposes a sequential approximate robust design optimization (SARDO) with the radial basis function (RBF) network. In RDO, mean and standard deviation of objective should be minimized simultaneously. Therefore, RDO is generally formulated as bi-objective optimization. Our goal to find optimal solution small number evaluations, not identifying set Pareto-optimal using Multi-Objective Evolutionary Algorithms. The weighted sum often used solution. contrast, lp norm method in this paper. Through illustrative examples, some validations are clarified. Next, SARDO RBF network discussed. general, functions obtained by finite difference method. Thus, order obtain functions, directly applied response surface. High accuracy will leads highly accurate avoid inaccurate numerical calculation, expressed only Gaussian kernel. As result, it expected that can found evaluations. validity proposed approach examined. Finally, variable blank holder force trajectory for reducing springback