이산분포의 AIC를 이용한 신뢰성 기반 최적설계

Abstract

Solutions to the deterministic design optimization cannot consider the uncertainties that may occur from the tolerance of design variables, noise and environmental factors, material properties, and uncontrollable operations. Therefore these solutions may cause the possibility of failure. However, reliability-based design optimization (RBDO) can provide the reliability of a system by mean of probabilistic design criteria, i.e., possibility of failure. To assure these criteria, many methods for the reliability analysis have been developed. However, these methods assume that data take one of continuous forms of distribution functions. In real problems, because real data may often provide discrete form, it is important to estimate the distributions from discrete data in the reliability analysis. In this paper, we introduce AIC method that can determine the best estimated distribution for discrete data. These distributions are used for reliability analysis in RBDO. Mathematical and engineering examples are illustrated to verify the proposed method.