A credible interval analysis method for uncertain structures under nonprobabilistic framework

Abstract

In this paper, a novel credible structural interval analysis method is proposed. Traditionally, nonprobabilistic uncertainties are modeled as interval or convex models, and no description of their credibility is given. Based on credible nonprobability uncertainty theory, the credible interval and ellipsoid are quantified to model irrelevant and relevant uncertain structural parameters. An efficient adaptive kriging (EAK) method is developed to solve the interval dynamic responses of structures with credible intervals and ellipsoids. In contrast to existing propagation methods, the proposed EAK method shares the samples of different time steps to improve the computational efficiency. A numerical example and two engineering applications are used to show the validity, accuracy and efficiency of the proposed method. Comparisons show that the proposed method can effectively calculate interval dynamic responses with credibility and is competitive when addressing nonlinear or coupled structural systems.