An innovative Bayesian updating method for laminated composite structures under evidence uncertainty

Abstract

Uncertainties widely exist in laminated composite structures and evidence theory is an effective tool to handle with these uncertainties. When new observations of the evidence variable are obtained, the corresponding representation model should be updated, but currently there is a lack of updating method for laminated composite structures under evidence uncertainty. Therefore, a Bayesian updating method is proposed to update the evidence model of laminated composite structures by new observations. First, the posterior basic probability assignment (BPA) is proposed as a key metric in Bayesian updating under evidence uncertainty. Second, a general Monte Carlo simulation (MCS) method is put forward to estimate posterior BPA. To improve the computational efficiency of the MCS method, two Kriging model based methods are subsequently developed for the Bayesian updating under evidence uncertainty by using different proxy strategies. Finally, the effectiveness of the proposed methods are demonstrated by two numerical examples and two laminated composite structures including a T300/QY8911 composite plate and wing structure with composite skin. The results show the proposed Bayesian updating can reasonably quantify the effect of new observations on the BPA for laminated composite structures under evidence uncertainty, and the proposed Kriging model based methods are efficient under ensuring the accuracy.