Reliability-based design optimization using adaptive surrogate model and importance sampling-based modified SORA method

Abstract

Reliability-based design optimization (RBDO) has been an important research field with the increasing demand for product reliability in practical applications. This paper presents a new RBDO method combining adaptive surrogate model and Importance Sampling-based Modified Sequential Optimization and Reliability Assessment (IS-based modified SORA) method, which aims to reduce the number of calls to the expensive objective function and constraint functions in RBDO. The proposed method consists of three key stages. First, the samples are sequentially selected to construct Kriging models with high classification accuracy for each constraint function. Second, the samples are obtained by Markov Chain Monte Carlo in the safety domain of design space. Then, another Kriging model for the objective function is sequentially constructed by adding suitable samples to update the Design of Experiment (DoE) of the objective function. Third, the expensive objective and constraint functions of the original optimization problem are replaced by the surrogate models. Then, the IS-based modified SORA method is performed to decouple reliability optimization problem into a series of deterministic optimization problems that are solved by a Genetic Algorithm. Several examples are adopted to verify the proposed method. The optimization results show that the proposed method can reduce the number of calls to the original objective function and constraint functions without loss of precision compared to the alternative methods, which illustrates the efficiency and accuracy of the proposed method.