Structural optimization based on reliability index mapping functions

摘要

Reliability Based Optimization (RBO) is often computationally intensive because the coupling of reliability evaluation and cost minimization. The mean first order reliability method (MFORM) is efficient for reliability evaluation, but it may not accurate. Nevertheless, the accurate reliability index is often highly correlated with the MFORM reliability index. In this study, we suggest that one can determine the relationship between the MFORM reliability index and the accurate reliability index through a regression analysis, which is called the reliability index mapping function. Based on the reliability index mapping function, an equivalent MFORM target reliability index can be determined such that MFORM can be used for reliability evaluation in RBO. Since the computational work involved in MFORM is minimal, RBO based on reliability index mapping functions is efficient. The reliability index mapping function determined via a regression analysis may only reflect the approximate relationship between the accurate reliability index and the MFORM reliability index. How to consider uncertainties in reliability index mapping functions in RBO is also described in this paper.