ABSTRACT
In structural engineering, the epistemic uncertainties of random variables, including distribution parameters and distribution type uncertainties, usually exist due to initial data scarcity. Moreover, the data will be updated over time, which causes the structural reliability updating. In this paper, for the initial data scarcity and successive data updating cases, a random variable model is developed to quantify the uncertainties of distribution parameter and distribution type based on the Bootstrap method, Akaike information criterion and Bayes update method. Considering the epistemic uncertainties, the reliability index of structure becomes a random variable and its probability density function (PDF) is presented. The numerical results show that the variability of the distribution parameters gradually decrease as the data update, and so does the variability of the reliability index. The PDF of the reliability index with the uncertainties of both distribution parameters and distribution type is largely different from that with only the uncertainty of distribution parameters.
