Competitive comparison of optimal designs of experiments for sampling-based sensitivity analysis

Abstract

Nowadays, the numerical models of real-world structures are more precise, more complex and, of course, more time-consuming. Despite the growth of a computational performance, the exploration of model behaviour remains a complex task. The sensitivity analysis is a basic tool for investigating the sensitivity of the model to its inputs. One widely used strategy to assess the sensitivity is based on a finite set of simulations for a given sets of input parameters, i.e. points in the design space. An estimate of the sensitivity can be then obtained by computing correlations between the input parameters and the chosen response of the model. The accuracy of the sensitivity prediction depends on the choice of design points called the design of experiments. The aim of the presented paper is to review and compare available criteria for the assessment of design of experiments suitable for sampling-based sensitivity analysis.

Bayesian parameter identification of lattice discrete particle model for concrete

“…In order to verify the pro-posed identification method, the synthetic experimental data are used in a form of 10 data sets with three repetitions of each test. The corresponding design of experiments is obtained by a stratified procedure called Latin hypercube sampling (LHS) which is able to respect the prescribed probability distributions (Janouchová and Kučerová (2013)). For convenience and readability, data are presented in terms of nominal stress σ N and nominal strain ε N .…”

Artificial neural networks in the calibration of nonlinear mechanical models

“…To achieve some reliable information from sensitivity analysis as well as a good approximation by an ANN, one has to choose the training data carefully according to a suitable design of experiments, see e.g. [38] for a competitive comparison of several experimental designs.…”

Simulation data management for adaptive design of experiments: A litterature review

“…However, the choice of the criterion is objective-dependent [1] proposed also to develop criteria for optimal design, as combination of a criterion to "identify the design region in which system performance is optimised" and design criteria "on the prediction error of the true output". Other criteria for this kind of DoE were presented in [43]. The authors give an assessment of presented criteria for sensitivity analysis.…”