Certified Metamodels for Sensitivity Indices Estimation
نویسندگان
چکیده
Global sensitivity analysis of a numerical code, more specifically estimation of Sobol indices associated with input variables, generally requires a large number of model runs. When those demand too much computation time, it is necessary to use a reduced model (metamodel) to perform sensitivity analysis, whose outputs are numerically close to the ones of the original model, while being much faster to run. In this case, estimated indices are subject to two kinds of errors: sampling error, caused by the computation of the integrals appearing in the definition of the Sobol indices by a Monte-Carlo method, and metamodel error, caused by the replacement of the original model by the metamodel. In cases where we have certified bounds for the metamodel error, we propose a method to quantify both types of error, and we compute confidence intervals for first-order Sobol indices. Résumé. L’analyse de sensibilité globale d’un modèle numérique, plus précisément l’estimation des indices de Sobol associés aux variables d’entrée, nécessite généralement un nombre important d’exécutions du modèle à analyser. Lorsque celles-ci requièrent un temps de calcul important, il est judicieux d’effectuer l’analyse de sensibilité sur un modèle réduit (ou métamodèle), fournissant des sorties numériquement proches du modèle original mais pour un coût nettement inférieur. Les indices estimés sont alors entâchés de deux sortes d’erreur: l’erreur d’échantillonnage, causée par l’estimation des intégrales définissant les indices de Sobol par une méthode de Monte-Carlo, et l’erreur de métamodèle, liée au remplacement du modèle original par le métamodèle. Lorsque nous disposons de bornes d’erreurs certifiées pour le métamodèle, nous proposons une méthode pour quantifier les deux types d’erreurs et fournir des intervalles de confiance pour les indices de Sobol du premier ordre.
منابع مشابه
Uncertainties Assessment in Global Sensitivity Indices Estimation from Metamodels
Global sensitivity analysis is often impracticable for complex and resource intensive numerical models, as it requires a large number of runs. The metamodel approach replaces the original model by an approximated code that is much faster to run. This paper deals with the information loss in the estimation of sensitivity indices due to the metamodel approximation. A method for providing a robust...
متن کاملEstimating Sensitivity Indices Based on Gaussian Process Metamodels with Compactly Supported Correlation Functions
Specific formulae are derived for quadrature-based estimators of global sensitivity indices when the unknown function can be modeled by a regression plus stationary Gaussian process using the Gaussian, Bohman, or cubic correlation functions. Estimation formulae are derived for the computation of process-based Bayesian and empirical Bayesian estimates of global sensitivity indices when the obser...
متن کاملSensitivity Analysis of Simulation Models
This contribution presents an overview of sensitivity analysis of simulation models, including the estimation of gradients. It covers classic designs and their corresponding (meta)models; namely, resolution-III designs including fractional-factorial twolevel designs for rst-order polynomial metamodels, resolution-IV and resolution-V designs for metamodels augmented with twofactor interactions,...
متن کاملGlobal sensitivity analysis via multi-fidelity polynomial chaos expansion
The presence of uncertainties are inevitable in engineering design and analysis, where failure in understanding their effects might lead to the structural or functional failure of the systems. The role of global sensitivity analysis in this aspect is to quantify and rank the effects of input random variables and their combinations to the variance of the random output. In problems where the use ...
متن کاملOrthogonal arrays for estimating global sensitivity indices of non-parametric models based on ANOVA high-dimensional model representation
Global sensitivity indices play important roles in global sensitivity analysis based on ANOVA high-dimensional model representation. However, few effective methods are available for the estimation of the indices when the objective function is a nonparametric model. In this paper, we explore the estimation of global sensitivity indices of non-parametric models. The main result (Theorem 2.1) show...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
عنوان ژورنال:
دوره شماره
صفحات -
تاریخ انتشار 2012