Nonparametric Bayesian stochastic model updating with hybrid uncertainties

Kitahara, Masaru; Bi, Sifeng; Broggi, Matteo; Beer, Michael

doi:10.1016/j.ymssp.2021.108195

Cited by 32 publications

(9 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Various techniques have been developed, e.g. staircase density function [38], Gaussian mixture model, and Bate mixture model [16], to parameterise the unknown distributions into undermined coefficients. This is actually a transformation of the distribution-free problem into the distribution-based problem by this additional parameterisation step.…”

Section: Perspectives and Future Challengesmentioning

confidence: 99%

Stochastic Model Updating with Uncertainty Quantification: An Overview and Tutorial

Beer

Cogan³

et al. 2023

Preprint

View full text Add to dashboard Cite

Section: Perspectives and Future Challengesmentioning

confidence: 99%

Stochastic Model Updating with Uncertainty Quantification: An Overview and Tutorial

Beer

Cogan³

et al. 2023

Preprint

View full text Add to dashboard Cite

“…In this work the comparison of the two datasets is performed based on Euclidean distance, however, it is pointed out, that different metrics can be employed as well, such as the Bhattacharyya distance, which is able to compare the statistical properties of the datasets [5]. In the following four Bayesian Updating methods are presented in more detail.…”

Section: Bayesian Model Updatingmentioning

confidence: 99%

“…These methods are widely employed and well established because of their potential, their fast and easy implementation and general applicability. Furthermore, they can be easily expanded by using them in combination with meta-modelling techniques or additional reliability methods such as Subset Simulation or Staircase Random Variables [4] [5]. At first Bayesian Model Updating and the four methods are introduced.…”

Section: Introductionmentioning

confidence: 99%

Comparison of state of the art sampling-based Bayesian Updating techniques

Dodt¹,

Kono²,

Broggi³

et al. 2022

8th International Symposium on Reliability Engineering and Risk Management

View full text Add to dashboard Cite

Nowadays, model updating has an increasing importance in many areas of interest for engineering applications such as structural health monitoring or risk and reliability assessment. As a matter of fact, it allows for solving a plethora of inverse problems with high computationally efficiency, allowing for example to monitor structural parameter, to detect damage quickly and to timely intervene with mitigating actions. Among the algorithms available to solve Bayesian updating, methods based on sampling present a flexibility that allows solving the problem numerically, either based on Markov Chain Monte Carlo (MCMC) method or based on the usage of Bayesian updating with structural reliability (BUS) methods. Additionally, BUS can be coupled with additional methods such as surrogate modelling and efficient simulation methods to further improve its numerical efficiency. Thus, engineering practitioners need to understand which possible combination of the available algorithms should be used to solve their needs. In this paper, we provide an overview of different MCMC and BUS methods, directly calling the model and also employing the Kriging meta-model, covering in detail the advantages and disadvantages of each method as well as their applicability. The investigated methods are applied to solve model updating and model class selection. Two numerical examples are used to verify and test the analysed methods, drawing conclusion on their accuracy, performance, robustness, numerical efficiency and ability to perform model class selection.

show abstract

“…We have recently developed a distributionfree approach to stochastic model updating, where the probabilistic model is defined using SDFs (Kitahara et al, 2022). This approach relies only on the bounded set of the probabilistic model, and within this set, a broad range of distributions are arbitrarily approximated using SDFs.…”

Section: Introductionmentioning

confidence: 99%

Stochastic Model Updating and Model Class Selection for Quantification of Different Types of Uncertainties

Kitahara,

Beer

2023

Proceeding of the 33rd European Safety and Reliability Conference

View full text Add to dashboard Cite

Stochastic model updating has been increasingly utilized in various engineering applications to quantify parameter uncertainty from multiple measurement datasets. We have recently developed a stochastic updating framework, in which the parameter distributions are approximated by staircase density functions (SDFs). This framework is applicable without any prior knowledge of the distribution formats; thus, it can be considered as a distribution-free approach. On the other hand, measurement uncertainty should also be considered in model updating since the measurement is typically performed under hard-to-control randomness. However, in model updating, it is difficult to distinguish different types of uncertainties in the measurement datasets, and measurement uncertainty is often embedded in parameter uncertainty. To address this issue, this study employs the Bayesian model class selection framework, in which different types of probabilistic models are used to represent different types of uncertainties and the most appropriate model is determined based on the associated evidence. In this sense, the proposed framework does not require any prior knowledge about the sources of uncertainty in the measurement datasets. Simple numerical examples are used to demonstrate the proposed framework.

show abstract

Nonparametric Bayesian stochastic model updating with hybrid uncertainties

Cited by 32 publications

References 23 publications

Stochastic Model Updating with Uncertainty Quantification: An Overview and Tutorial

Stochastic Model Updating with Uncertainty Quantification: An Overview and Tutorial

Comparison of state of the art sampling-based Bayesian Updating techniques

Stochastic Model Updating and Model Class Selection for Quantification of Different Types of Uncertainties

Contact Info

Product

Resources

About