An efficient dimensionality-independent algorithm for failure probability-based global sensitivity analysis by dual-stage adaptive kriging model

Yun, Wanying; Lü, Zhenzhou; Jiang, Xian; He, Peter

doi:10.1080/0305215x.2020.1814273

Cited by 5 publications

(5 citation statements)

References 41 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As shown in Figure 1, the EAK modeling is mainly divided into two stages: in Stage I , the initial Kriging model with a small number of sample points is first constructed and the initial state of the Markov chain is determined; considering the MCMC algorithm, the samples in failure domain are identified; by performing the K -means cluster analysis (Yun et al ., 2020; Feng et al ., 2020) for the samples in failure domain, the IS samples are generated by centering on the obtained K center-of-mass points, respectively. In Stage II , regarding the IS samples in the reduced sampling pool as candidate samples, updating the active Kriging model until the convergence criterion is satisfied.…”

Section: Hierarchical Collaborative Enhanced Active Krigingmentioning

confidence: 99%

Fatigue reliability framework using enhanced active Kriging-based hierarchical collaborative strategy

Zhang

Song

Bai

2023

IJSI

View full text Add to dashboard Cite

PurposeThe purpose of this study is to improve the computational efficiency and accuracy of fatigue reliability analysis.Design/methodology/approachBy absorbing the advantages of Markov chain and active Kriging model into the hierarchical collaborative strategy, an enhanced active Kriging-based hierarchical collaborative model (DCEAK) is proposed.FindingsThe analysis results show that the proposed DCEAK method holds high accuracy and efficiency in dealing with fatigue reliability analysis with high nonlinearity and small failure probability.Research limitations/implicationsThe effectiveness of the presented method in more complex reliability analysis problems (i.e. noisy problems, high-dimensional issues etc.) should be further validated.Practical implicationsThe current efforts can provide a feasible way to analyze the reliability performance and identify the sensitive variables in aeroengine mechanisms.Originality/valueTo improve the computational efficiency and accuracy of fatigue reliability analysis, an enhanced active DCEAK is proposed and the corresponding fatigue reliability framework is established for the first time.

show abstract

Section: Hierarchical Collaborative Enhanced Active Krigingmentioning

confidence: 99%

Fatigue reliability framework using enhanced active Kriging-based hierarchical collaborative strategy

Zhang

Song

Bai

2023

IJSI

View full text Add to dashboard Cite

show abstract

“…As shown in Figure 2, for a BN with three root nodes {𝑋 1 , 𝑋 2 , 𝑋 3 }, two intermediate nodes {𝑌 1 , 𝑌 2 }, and one leaf node 𝑆. According to Equation (2), the joint probability distribution of all nodes according to the chain rule of joint probability distribution can be expressed as Equation (3).…”

Section: Bn Inferencementioning

confidence: 99%

“…On the other hand, uncertainty information affecting the reliability of complex systems is more and more diverse due to multiple factors such as incomplete experiments, design flaws, processing errors, cognitive limitations, service environments, and so forth 2 . Therefore, how to effectively quantify system uncertainties when performing reliability analysis of systems is likewise one of the main issues in the field of reliability analysis 3 …”

Section: Introductionmentioning

confidence: 99%

“…2 Therefore, how to effectively quantify system uncertainties when performing reliability analysis of systems is likewise one of the main issues in the field of reliability analysis. 3 Uncertainties have been classified into two categories: one is inherent and irreducible uncertainty (also called random uncertainty or aleatory uncertainty), which can be described and quantified by probability theory; the other is cognitive or epistemic uncertainty caused by the lack of knowledge and data. Many theories have been developed to quantify the epistemic uncertainty, including evidence theory, 4 fuzzy theory, 5 probability box (P-box), 6 random set theory, 7 Bayesian theory, 8 possibility theory, 9 and so forth.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Reliability analysis of systems with mixed uncertainties based on Bhattacharyya distance and Bayesian network

Jia

Sun

et al. 2022

Quality & Reliability Eng

View full text Add to dashboard Cite

The lack of data due to the high cost of experiments, limited knowledge of system structure, and coupling failure mechanism make it necessary to consider uncertainties when analyzing the reliability of complex systems. This paper provides a comprehensive work for system reliability analysis under consideration of mixed uncertainties. First, Bayesian networks (BNs) are used to model the structure of complex systems. Second, mixed uncertainties are described via a probability box and accurately quantified by the Bhattacharyya distance, which can synthetically represent the mean and variance of two data sets. Then, the importance of components with mixed uncertainties is defined and analyzed. Finally, a comprehensive sensitivity index is proposed for sensitivity analysis and epistemic uncertainty tracing of data sources based on the Bhattacharyya distance. This method is used to analyze the reliability of an auxiliary power supply system of a train and verifies the feasibility and accuracy of the proposed method.

show abstract

“…The same indices can be obtained using the contrast function [13] where Pf is a minimizer. The rationality of sensitivity indices [11][12][13] has been shown in many applications [16][17][18][19][20][21][22][23][24][25].…”

Section: Introductionmentioning

confidence: 99%

New Importance Measures Based on Failure Probability in Global Sensitivity Analysis of Reliability

Kala

2021

Mathematics

View full text Add to dashboard Cite

This article presents new sensitivity measures in reliability-oriented global sensitivity analysis. The obtained results show that the contrast and the newly proposed sensitivity measures (entropy and two others) effectively describe the influence of input random variables on the probability of failure Pf. The contrast sensitivity measure builds on Sobol, using the variance of the binary outcome as either a success (0) or a failure (1). In Bernoulli distribution, variance Pf(1-Pf) and discrete entropy—Pfln(Pf)-(1-Pf)ln(1-Pf) are similar to dome functions. By replacing the variance with discrete entropy, a new alternative sensitivity measure is obtained, and then two additional new alternative measures are derived. It is shown that the desired property of all the measures is a dome shape; the rise is not important. Although the decomposition of sensitivity indices with alternative measures is not proven, the case studies suggest a rationale structure of all the indices in the sensitivity analysis of small Pf. The sensitivity ranking of input variables based on the total indices is approximately the same, but the proportions of the first-order and the higher-order indices are very different. Discrete entropy gives significantly higher proportions of first-order sensitivity indices than the other sensitivity measures, presenting entropy as an interesting new sensitivity measure of engineering reliability.

show abstract

An efficient dimensionality-independent algorithm for failure probability-based global sensitivity analysis by dual-stage adaptive kriging model

Cited by 5 publications

References 41 publications

Fatigue reliability framework using enhanced active Kriging-based hierarchical collaborative strategy

Fatigue reliability framework using enhanced active Kriging-based hierarchical collaborative strategy

Reliability analysis of systems with mixed uncertainties based on Bhattacharyya distance and Bayesian network

New Importance Measures Based on Failure Probability in Global Sensitivity Analysis of Reliability

Contact Info

Product

Resources

About