Efficient adaptive Kriging-based reliability analysis combining new learning function and error-based stopping criterion

Yi, Jin; Zhou, Qi; Cheng, Yuansheng; Liu, Jun

doi:10.1007/s00158-020-02622-3

Cited by 43 publications

(21 citation statements)

References 52 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Dimension reduction was applied to reduce the variables based on the sensitivity analysis results presented in Table 2, considering the ''dimension curse'' in the reliability analysis of high dimension problems. 30 Therefore, four variables presented in Table 3, X 1 , X 2 , X 8 , and X 9 , were regarded as the mainly input variables to construct the reliability analysis model, as shown in equation ( 9), considering the predictive power of the KG model for high dimensional functions.…”

Section: Sensitivity Analysismentioning

confidence: 99%

Sealing reliability assessment of deep-water oil and nature gas pipeline connector considering thermo-mechanical coupling

Zeng

Sun

Zhao

et al. 2021

Proceedings of the Institution of Mechanical Engineers, Part M:

View full text Add to dashboard Cite

The deep-water pipeline connector (DPC) is a key piece of equipment for the connection of multiple facilities located in different positions of the subsea production system. Its sealing reliability affects the safety of the production system directly during oil and gas exploration. Although the reliability assessment methods of the DPC have been investigated, previous studies have mostly been carried out under a single mechanical field. To determine the DPC reliability considering thermo-mechanical coupling, this paper establishes random load models of working pressure and temperature by using the Gaussian function, and the stochastic finite element model of the DPC is constructed to obtain the sealing performance under coupling conditions. A Kriging–Sobol sensitivity analysis method is developed to obtain the main variables influencing the sealing performance for the purpose of decreasing the computational cost of the ordinary sensitivity analysis. A reliability assessment approach based on the cross entropy-importance sampling-Kriging method is introduced to analyze the DPC reliability, and a sealing reliability assessment method of DPC under thermo-mechanical coupling conditions is finally formed. An engineering case is taken to verify the effectiveness of the proposed method. The results show that the reliability analysis accuracy of the proposed method is almost agreed with the Monte Carlo method, but the computational cost can be reduced 85.61%, which indicates that the proposed method provides designers with a fast method with an acceptable computational cost to assess the reliability of the DPC connector under coupling conditions.

show abstract

Section: Sensitivity Analysismentioning

confidence: 99%

Sealing reliability assessment of deep-water oil and nature gas pipeline connector considering thermo-mechanical coupling

Zeng

Sun

Zhao

et al. 2021

Proceedings of the Institution of Mechanical Engineers, Part M:

View full text Add to dashboard Cite

show abstract

“…Model-specific criteria include those that require ensembles or a Bayesian model (Olsson and Tomanek, 2009;Hino, 2020, 2021). Domain-specific criteria include work on drug target prediction, automated screening in systematic reviews, and a larger body of work focused on structural reliability analysis (Temerinac-Ott et al, 2015;Callaghan and Müller-Hansen, 2020;Moustapha et al, 2021;Gaspar et al, 2015;Yi et al, 2020).…”

Section: Related Workmentioning

confidence: 99%

Hitting the Target: Stopping Active Learning at the Cost-Based Optimum

Pullar-Strecker¹,

Dost²,

Frank³

et al. 2021

Preprint

View full text Add to dashboard Cite

Active learning allows machine learning models to be trained using fewer labels while retaining similar performance to traditional fully supervised learning. An active learner selects the most informative data points, requests their labels, and retrains itself. While this approach is promising, it leaves an open problem of how to determine when the model is 'good enough' without the additional labels required for traditional evaluation. In the past, different stopping criteria have been proposed aiming to identify the optimal stopping point. However, optimality can only be expressed as a domain-dependent trade-off between accuracy and the number of labels, and no criterion is superior in all applications. This paper is the first to give actionable advice to practitioners on what stopping criteria they should use in a given real-world scenario. We contribute the first large-scale comparison of stopping criteria, using a cost measure to quantify the accuracy/label trade-off, public implementations of all stopping criteria we evaluate, and an open-source framework for evaluating stopping criteria. Our research enables practitioners to substantially reduce labelling costs by utilizing the stopping criterion which best suits their domain.

show abstract

“…• New stopping criteria: New stopping criteria are developed for more accurate termination of the learning process to avoid premature of the algorithm or unnecessary calls to the performance functions, and the include the error-based stopping criterion (ESC) [74,75] and its improvement based on bootstrap confidence estimation (BCE) [76].…”

Section: Introductionmentioning

confidence: 99%

“…In this way, the computational efficiency of ABSVR can be enhanced by using a set of important samples. Moreover, a hybrid stopping criterion based on the bootstrap confidence estimation (BCE) proposed in [76] is developed to terminate the active learning process, ensuring that the learning algorithm stops at an appropriate stage. The proposed ABSVR is easy to implement since no embedded optimization algorithm nor iso-probabilistic transformation is required.…”

Section: Introductionmentioning

confidence: 99%

Efficient structural reliability analysis based on adaptive Bayesian support vector regression

Wang

et al. 2021

Computer Methods in Applied Mechanics and Engineering

View full text Add to dashboard Cite

Efficient adaptive Kriging-based reliability analysis combining new learning function and error-based stopping criterion

Cited by 43 publications

References 52 publications

Sealing reliability assessment of deep-water oil and nature gas pipeline connector considering thermo-mechanical coupling

Sealing reliability assessment of deep-water oil and nature gas pipeline connector considering thermo-mechanical coupling

Hitting the Target: Stopping Active Learning at the Cost-Based Optimum

Efficient structural reliability analysis based on adaptive Bayesian support vector regression

Contact Info

Product

Resources

About