An efficient polynomial chaos-enhanced radial basis function approach for reliability-based design optimization

Reliability-based design optimization (RBDO) has been widely used to search for the optimal design under the presence of parameter uncertainty in the engineering application. Unlike traditional deterministic optimization, RBDO problem takes the uncertainty of design variables and probabilistic reliability constraints into consideration. In the context of RBDO, a large number of model evaluations are required in the reliability analysis to estimate the failure probability. However, the intensive computation of reliability analysis makes it infeasible to address complex and expensive problems. In order to relieve the computational burden, an efficient polynomial chaos-enhanced radial basis function (PCE-RBF) approach is proposed. In this approach, RBF combined with sparse PC method is constructed to enhance predictive accuracy of metamodel. To refine the metamodel, local variation with minimum distance sampling criterion is proposed to select the sample points sequentially. Then, the refined PCE-RBF metamodel with acceptable accuracy is used to perform gradient-based optimization for solving RBDO problem. The performance of the proposed method is validated by four benchmark examples and truss structure issue.