.png){kind=icon}
返回
A novel single-loop estimation method for predictive failure probability-based global sensitivity analysis
.png){kind=icon}
PRE
AI
DOI：10.1016/j.ress.2024.110249.png)
.png){kind=icon}
原文求助.png){kind=icon}
分享.png){kind=icon}
收藏
摘要
En 中文
The existing predictive failure probability (PFP)-based global sensitivity analysis methods may get stuck in practical application due to the large amount of calculation or insufficient precision. Therefore, a novel singleloop estimation method is proposed in this paper to effectively estimate PFP-based global sensitivity measures. Firstly, by introducing standard normal auxiliary variables and equivalent probability transformation, the overall uncertainty of each input variable is decomposed into its inherent uncertainty and distribution parameter uncertainty. Then, Bayes theorem -based formulations are derived for three PFP-based global sensitivity measures, where only the unconditional PFP and the conditional marginal and joint distributions of standard normal auxiliary variables and distribution parameters need to be known. These items can be evaluated simultaneously using only one set of functional calls. Finally, adaptive Kriging model is embedded into the proposed method in place of the time -demanding performance function to further enhance computational efficiency. The effectiveness of the proposed methods is validated through four benchmark reliability problems asnd one engineering problem.
Keyword:
Predictive failure probability
Global sensitivity analysis
Single -loop estimation
Bayes theorem
Adaptive kriging model
期刊
R
IF:
11
论文数:
9.0K
被引数:
4.2W
机构
引用论文
.png){kind=icon}
分享收藏
查看更多内容