Stochastic Finite Element Model Updating of Aging Jacket Platform Based on Kriging Model and Inverse Kullback-Leibler Divergence

Aiming at the problems of limited measured data and poor updating effect of jacket platform, a two-stage stochastic model updating method based on inverse Kullaback-Leibler(KL) divergence is proposed in this paper: The constructing Kriging models as substitutes for finite element models in computations, the first stage is frequency distance updating; The second stage is frequency distribution difference updating. Firstly, assuming that the structural parameters to be updated and the modal parameters obey the Gaussian distribution, an updating problem of a model with uncertainty is transformed into the mean and standard deviation. Secondly, taking the jacket platform as an example, the correction effect of the proposed method when the mean and standard deviation of the modified parameters were unknown was studied, and the influence of different amounts of frequency data on the correction results was explored. Finally, the proposed method was applied to a jacket platform in Bohai Bay, The results show that the mean frequency correction error of the platform was less than 0.3%, and the standard deviation correction error was less than 7%, the proposed method can be applied to actual engineering.