A global nonprobabilistic reliability sensitivity analysis in the mixed aleatory-epistemic uncertain structures

The aim of this paper is to account for the effect of the epistemic uncertainty of the input variables' uncertainty in the nonprobabilistic reliability analysis on the safety of the structure system. Based on the idea of moment-independent sensitivity analysis, a modified sensitivity measure of the nonprobabilistic reliability is constructed to identify the most influential epistemic parameters of interval variables. For calculating the nonprobabilistic reliability sensitivity measures of the epistemic variables, a computational model is established. And a solution method with the advantages of the state-dependent parameter model is employed to improve the computational efficiency and avoid the complex sampling procedure. The numerical examples and engineering examples show that the proposed method of solving the sensitivity measure is reasonable and effective. The sensitivity measure of nonprobabilistic reliability proposed in this paper can give an essential importance sequence of all the epistemic uncertainties and identify key contributing epistemic uncertainties. When the sensitivity measure is larger, the epistemic uncertainty variable will become more important and should collect the data to increase knowledge of parameters. The sensitivity measures can provide the availability guidance to reduce the number of epistemic variables.