Fast Bayesian model selection with application to large locally-nonlinear dynamic systems

S. De, E. A. Johnson, S. F. Wojtkiewicz

Research output: Contribution to journalConference articlepeer-review

2 Scopus citations

Abstract

Bayesian model selection chooses, based on measured data, using Bayes' theorem, suitable mathematical models from a set of possible models. In structural analysis, linear models are often used to facilitate design and analysis, though they do not always accurately reproduce actual structural responses. When the models also require the inclusion of nonlinearity to improve accuracy, the computation time required for response simulation increases significantly. To reduce this computational burden, this paper proposes incorporating into the model selection process an efficient dynamic response algorithm previously developed by the last two authors for locally nonlinear systems. Additionally, nested sampling, an intelligent sampling algorithm, is used to reduce the number of simulations (using whichever response simulation algorithm) needed for accurate posterior distribution computation. A numerical example of a 20-story three-dimensional building with roof-mounted tuned mass dampers (TMDs), using different linear and nonlinear damping models as the candidates to reproduce the TMD damping, demonstrates that the proposed approach is up to 1000 times faster than traditional Bayesian model selection employing a conventional structural response solver.

Keywords

  • Bayesian model selection
  • Nested sampling
  • Nonlinear Volterra integral equation

ASJC Scopus subject areas

  • Civil and Structural Engineering
  • Building and Construction

Fingerprint

Dive into the research topics of 'Fast Bayesian model selection with application to large locally-nonlinear dynamic systems'. Together they form a unique fingerprint.

Cite this