.png){kind=icon}
返回
Enhancing Structural Reliability through AI-Driven Control Variates and Subset Simulation
.png){kind=icon}
PRE
AI
DOI：10.1061/AJRUA6.RUENG-1619.png)
.png){kind=icon}
原文求助.png){kind=icon}
分享.png){kind=icon}
收藏
摘要
En 中文
In structural reliability analysis for engineering systems with time-consuming and complicated models, it is a crucial task to estimate the accurate failure probability with a minimum number of simulations. In order to further improve the efficiency of structural reliability analysis for costly problems, this study introduces, develops, and applies a new metamodel combining artificial intelligence (AI)-driven control variates (CVs) and subset simulation (SS) techniques. To meet this aim, the design of experiments (DoEs) is first generated by increasing the standard deviation of variables, and the failure event is transformed into intermediate failure events using the SS strategy. Then, an AI-based method is constructed from the DoEs and is continuously updated in each intermediate failure event using the samples in the failure zone, allowing for the estimation of the true limit-state function (LSF) without direct computation. Finally, the CV approach is employed to refine the estimated failure probability. To validate the applicability and efficiency of the proposed AI-driven simulation approach, several challenging reliability problems, including the numerical and engineering examples, are solved. The results demonstrate that the proposed AI-driven simulation framework in this study is highly promising dealing with linear, nonlinear, and complicated LSFs, as well as extensive engineering problems.
Keyword:
Structural reliability
Artificial intelligence
Control variate
Metamodels
Subset simulation
Expensive problems
期刊
机构
引用论文
.png){kind=icon}
分享收藏
查看更多内容