Research of Physical Parameter Identification and Damage Localization based on the Gibbs Sampling

摘要

A new method for structural physical parameter identification is proposed for linear structure.Firstly,a linear structural identification model was obtained based on a series of transformation of the dynamic characteristic equation.Then the posterior distribution of the model is obtained by the Bayesian updating theory.Using the structural modal parameters and considering their randomness,the structural stiffness parameter is obtained from the conditional posterior distribution of the linear structural identification model.The Gibbs sampling based on the Markov Chain Monte Carlo(MCMC)method is employed during the process.In order to illustrate the proposed method,a 3-DOF linear shear building is used as an example to detect and quantify its damage based on model data measured before and after a severe loading event.The research shows that damage level and locations can be identified with little error by using proposed method.