Lye, Adolphus 
ORCID: 0000-0002-1803-8344, Cicirello, Alice and Patelli, Edoardo ORCID: 0000-0002-5007-7247
(2020)
Bayesian Model Updating of Reliability Parameters using Transitional Markov Chain Monte Carlo with Slice Sampling.
In: Proceedings of the 29th European Safety and Reliability Conference (ESREL), 2020-11-1 - 2020-11-5, Venice.
|
Text
ESREL 2020 Conference Paper (Final submission).pdf - Author Accepted Manuscript Download (925kB) | Preview |
Abstract
This research work presents a comparison of the performances between the Transitional Markov Chain Monte Carlo (TMCMC) and the TMCMC-Slice algorithm. Transitional Markov Chain Monte Carlo (TMCMC) algorithm is a popular approach in the estimation of epistemic parameters from Bayesian Inference. By sampling from a series of intermediate probability density functions, the sampler can generate samples from any target probability density functions. In the TMCMC algorithm, the Metropolis-Hastings sampling algorithm is adopted to generate samples from the intermediate probability density functions whilst in TMCMC-Slice algorithm, the Slice sampling algorithm is adopted to do so. In this work, the performance of the TMCMC-Slice over the TMCMC sampler is investigated for different number of samples. For this purpose, the two samplers are then adopted in a reliability parameter update of the Emergency Diesel Generator system that is employed in Daya Bay Nuclear Power Plant. The results show that while the TMCMC-Slice approach is able to produce slightly more precise estimates compared to the TMCMC approach, the computational time evolved in the case of the former was significantly greater compared to the latter. In addition, the Two-sample Kolmogorov–Smirnov test also provided sufficient evidence to reject the null hypothesis that the samples obtained from both techniques are from the same distribution at 5% level of significance.
| Item Type: | Conference or Workshop Item (Unspecified) |
|---|---|
| Uncontrolled Keywords: | Nuclear Power Plant, Reliability, Bayesian Model Updating, Slice Sampling, Transitional Markov Chain Monte Carlo |
| Depositing User: | Symplectic Admin |
| Date Deposited: | 10 Nov 2020 09:56 |
| Last Modified: | 18 Jan 2023 23:23 |
| DOI: | 10.3850/978-981-14-8593-0_4374-cd |
| Related URLs: | |
| URI: | https://livrepository.liverpool.ac.uk/id/eprint/3106185 |
Dimensions
Dimensions
