An separation of the aleatory and epistemic uncertainties with correlated normal variables and analysis of the influence of parameters

For the structural system with correlated variables with aleatory and epistemic uncertainties, the influence of distribution parameters with epistemic uncertainty to the output is studied. Firstly, orthogonal transformation is used to separate the aleatory and epistemic uncertainties. The epistemic uncertainty is separated from the aleatory uncertainty of input variables through the auxiliary variable. After that, the main effect of distribution parameters to the model output is discussed. Secondly, the equality between the first order variance of parameters to the expectation of the output and the first order variance contribution of parameters to the output after separating uncertainties is validated. At last, some examples are utilized to demonstrate the rationality and efficiency of method used to separate the aleatory and epistemic uncertainties.