A single-loop reliability-based design optimization using adaptive differential evolution

Reliability-based design optimization (RBDO) is an efficient tool for generating reliable and optimal solution under uncertainty of design variables. The major challenges in solving RBDO problems are generating optimal or near optimal reliable solution and higher computational cost. In this paper, a single-loop RBDO formulation is developed for addressing these challenges that uses shifting vector approach for achieving feasibility for violated constraints or performance functions. The formulation also incorporates target and trial vectors of differential evolution (DE) for guiding the algorithm. DE is also made adaptive by designing a heuristic parameter that controls two mutation operators for both exploration and exploitation of search space. The proposed RBDO method is tested on three mathematical and four engineering examples. The reliability of obtained solutions from RBDO methods are verified using Monte Carlo simulations with sample size of one million. The results of the proposed method are compared with various RBDO methods from the literature and a double-loop based DE method. It is found that the proposed method generates the best reliable solution for all examples. (c) 2022 Elsevier B.V. All rights reserved.

Automated summary

Reliability-based design optimization (RBDO) is an efficient tool for generating reliable and optimal solutions under uncertainty in design variables. This paper proposes a single-loop RBDO formulation that addresses the challenges of generating optimal reliable solutions and higher computational costs. The formulation incorporates a shifting vector approach and utilizes target and trial vectors of differential evolution (DE) to guide the algorithm. The proposed RBDO method is tested on mathematical and engineering examples, and its reliability is verified using Monte Carlo simulations.