Damage assessment research on frame structure based on Bayesian statistical inference

Basic theory of Bayesian statistics is introduced in this paper, in order to solve the difficulties of the denominator hardly in integral due to the larger dimensions, the method of Markov chain's Monte Carlo (MCMC) simulation is introduced, the samples in multi-dimension space are produced by prior probability, the damage in structures can be deduced by posterior distribution. The modal experiments were done on a frame structure by enhancing the local column on soil foundation in the laboratory before damage and after damage, the modal parameters were obtained. The modal parameters in non-damage status are used to obtain the mean value of posterior distribution by Bayesian inference, and it is used as the first step of model updating. In the second step, the mean value of the former distribution has been used to update the model, and then the posterior distribution is recalculated. The mean value obtained in step two can identify the location of the damage. The identification results also show that the foundation has important influence on the identification results.