Parsimonious Estimation of Probabilities of Undesirable Events in Structural Reliability Frameworks
We consider the exceedance of a threshold reliability value by the unidimensional output of a computer code G with multivariate probabilistic input v. When G is assumed monotonous with respect to v, the Monotonous Reliability Method, proposed by de Rocquigny (2009) in an engineering context, can provide bounds and crude estimates of the probability of occurence pf of such an undesirable event, using a sequential design of experiments. In the talk associated to this article, the formalization and a technical deepening of this idea are considered, as a large basis for future theoretical and applied studies in structural reliability. Two refined statistical estimators of pf are defined and their parsimonious properties are showed, which makes them useful in the frequent industrial context when the code G is time-consuming.
Structural reliability exceedance probability computer code input uncertainty Monte Carlo acceleration
NICOLAS BOUSQUET
EDF R&D, MRJ Department, 6 quai Watier, 78401 Chatou, Prance
国际会议
北京
英文
566-570
2011-06-20（万方平台首次上网日期，不代表论文的发表时间）