Research on reliability sensitivity for ranking the importance of random influential factors

摘要

During the reliability analysis, the influential factors that play important role on reliability, mainly contain two aspects: the influence of the randomicity of the influential factors itself on reliability and its coupling influence combined with the randomicity of other influential factors on reliability. Considering the above two aspects of influence, two kinds of reliability sensitivity, including single reliability sensitivity and synthetical reliability sensitivity, are defined in this paper for estimating the influence of each variable on reliability. To improve the computation efficiency of the above methods and expand their applicability, especially aiming at solving the case that the limit state function is implicit and its solution consumes much time of computer simulation, a fast method for computing reliability sensitivity by combining the kriging model with Monte Carlo simulation is presented in this paper. As shown by the example, agreement between results computed by both methods appears very good and the proposed methods are efficient and available for engineering application.