Bayesian model updating of a coupled-slab system using field test data utilizing an enhanced Markov chain Monte Carlo simulation algorithm. (1st November 2015)

• Record Type: Journal Article
• Authors: Lam, Heung-Fai; Yang, Jiahua; Au, Siu-Kui
• Is Part Of: Engineering structures. Volume 102(2015:Nov. 01)
• Page Start: 144
• Page End: 155
• Publication Date: 2015-11-01
• Subjects: Bayesian model updating -- Markov chain Monte Carlo simulation -- Model class -- Field test; Structural engineering -- Periodicals; Structural analysis (Engineering) -- Periodicals; Construction, Technique de la -- Périodiques; Génie parasismique -- Périodiques; Pression du vent -- Périodiques; Earthquake engineering; Structural engineering; Wind-pressure; Periodicals; 624.105
• DOI: 10.1016/j.engstruct.2015.08.005
• Languages: English
• ISSNs: 0141-0296

Abstract

Highlights: Model updating of an existing structural system via MCMC based on field test data. Finite element modeling of the targeted coupled-slab structural system using ANSYS. The enhancement of the MCMC algorithm by a new stopping criteria. The approximation of the posterior PDF based on the MCMC samples. Effects of the complexity of the model class in the posterior uncertainties. Abstract: Markov chain Monte Carlo (MCMC) simulation is applied for model updating of the coupled-slab system of a building structure based on field test data following the Bayesian theory. It is found that the identifiability of the model updating problem depends very much on the complexity of the class of models. By MCMC, the same algorithm can be used no matter the model updating problem is locally identifiable or not. The posterior joint probability density function (PDF) of model parameters is derived with consideration of the uncertainties from both the measurement noise and modeling error. To obtain a posterior PDF that is not analytically available in the complicated parameter space, an MCMC algorithm is proposed to sample a set of models in high-probability regions for the representation (or approximation) of the posterior PDF. The sampling process is divided into multiple levels, and individual bridge PDFs are constructed at each level that finally converged to the target posterior PDF. The samples move smoothly through each level and finally arrive at the important region of the target posterior PDF. A novel stopping criterion for the MCMC algorithm is proposed from the insight of the derivation of the posterior PDF. In the field test verification, the posterior marginal PDFs conditional on two model classes are obtained by the proposed MCMC algorithm, which provide valuable information about the identifiability of different model parameters.