Reliability-based design optimization under sampling uncertainty: shifting design versus shaping uncertainty
| Autor: | Sangjune Bae, Nam H. Kim, Seung-gyo Jang |
|---|---|
| Rok vydání: | 2018 |
| Předmět: |
0209 industrial biotechnology
Mathematical optimization Control and Optimization media_common.quotation_subject 0211 other engineering and technologies Finite difference method Process (computing) Sampling (statistics) 02 engineering and technology Computer Graphics and Computer-Aided Design Computer Science Applications 020901 industrial engineering & automation Control and Systems Engineering Point (geometry) Sensitivity (control systems) Engineering design process Software Normality Reliability (statistics) 021106 design practice & management media_common Mathematics |
| Zdroj: | Structural and Multidisciplinary Optimization. 57:1845-1855 |
| ISSN: | 1615-1488 1615-147X |
| DOI: | 10.1007/s00158-018-1936-0 |
| Popis: | The objective of this paper is a tradeoff between changing design and controlling sampling uncertainty in reliability-based design optimization (RBDO). The former is referred to as ‘living with uncertainty’, while the latter is called ‘shaping uncertainty’. In RBDO, a conservative estimate of the failure probability is defined using the mean and the upper confidence limit, which are obtained from samples and from the normality assumption. Then, the sensitivity of the conservative probability of failure is derived with respect to design variables as well as the number of samples. It is shown that the proposed sensitivity is much more accurate than that of the finite difference method and close to the analytical sensitivity. A simple RBDO example showed that once the design variables reach near the optimum point, the number of samples is adjusted to satisfy the conservative reliability constraints. This example showed that not only shifting design but also shaping uncertainty plays a critical role in the optimization process. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|
Full text from SpringerLink