Gaussian process surrogates for failure detection: A Bayesian experimental design approach

JOURNAL OF COMPUTATIONAL PHYSICS

卷:313

Article

Wang, Hongqiao¹  Lin, Guang²  Li, Jinglai³ 

[1] Shanghai Jiao Tong Univ, Dept Math, Inst Nat Sci, Shanghai 200240, Peoples R China

[2] Purdue Univ, Sch Mech Engn, Dept Math, 150 N Univ St, W Lafayette, IN 47907 USA

[3] Shanghai Jiao Tong Univ, Dept Math, MOE Key Lab Sci & Engn Comp, Inst Nat Sci, Shanghai 200240, Peoples R China

关键词: Bayesian inference;    Experimental design;    Failure detection;    Gaussian processes;    Monte Carlo;    Response surfaces;    Uncertainty quantification;   

DOI  :  10.1016/j.jcp.2016.02.053

来源: Elsevier

【 摘 要 】

An important task of uncertainty quantification is to identify the probability of undesired events, in particular, system failures, caused by various sources of uncertainties. In this work we consider the construction of Gaussian process surrogates for failure detection and failure probability estimation. In particular, we consider the situation that the underlying computer models are extremely expensive, and in this setting, determining the sampling points in the state space is of essential importance. We formulate the problem as an optimal experimental design for Bayesian inferences of the limit state (i.e., the failure boundary) and propose an efficient numerical scheme to solve the resulting optimization problem. In particular, the proposed limit-state inference method is capable of determining multiple sampling points at a time, and thus it is well suited for problems where multiple computer simulations can be performed in parallel. The accuracy and performance of the proposed method is demonstrated by both academic and practical examples. (C) 2016 Elsevier Inc. All rights reserved.

【 授权许可】

Free