期刊
RELIABILITY ENGINEERING & SYSTEM SAFETY
卷 106, 期 -, 页码 86-93出版社
ELSEVIER SCI LTD
DOI: 10.1016/j.ress.2012.04.011
关键词
Adaptive metropolis; Finite element; Markov chain simulation; Reliability; Support vector density
资金
- National Natural Science Foundation of China [10902028, 50978078]
- Australian Research Council [DP110104263]
- fundamental research fund for the central universities of China [HIT.NSRIF.2010011]
An importance sampling method based on the adaptive Markov chain simulation and support vector density estimation is developed in this paper for efficient structural reliability assessment. The methodology involves the generation of samples that can adaptively populate the important region by the adaptive Metropolis algorithm, and the construction of importance sampling density by support vector density. The use of the adaptive Metropolis algorithm may effectively improve the convergence and stability of the classical Markov chain simulation. The support vector density can approximate the sampling density with fewer samples in comparison to the conventional kernel density estimation. The proposed importance sampling method can effectively reduce the number of structural analysis required for achieving a given accuracy. Examples involving both numerical and practical structural problems are given to illustrate the application and efficiency of the proposed methodology. (C) 2012 Elsevier Ltd. All rights reserved.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据