Journal Article

SBSC+SRU: an error-guided adaptive Kriging method for expensive system reliability analysis

Cheng, Yuansheng, Liu, Jun, Yi, Jiaxiang

2022

Overview

In this paper, SBSC+SRU: an error-guided adaptive Kriging modeling method is proposed for the system reliability analysis with multiple failure modes. Therein, the accuracies of Kriging models will be improved by a novel learning function, in which the magnitude of Component Limit State Functions (CLSFs), uncertainties of Kriging models, and the coupling relationships among CLSFs are considered to identify the location and component index of the new sample. Then, the maximum estimated relative error of predicted failure probability is derivated by quantifying the probability of wrong sign prediction of samples. To be specific, the highly uncertain samples are first defined, after that the probability of wrong sign prediction of each highly uncertain sample is deduced combining the predictions of Kriging models and coupling relationship among all CLSFs. Therefore, the proposed approach knows the real-time estimated error and could terminate the adaptive updating process under the accuracy requirement. Three numerical examples including parallel and series system problems and an engineering case concerning the system reliability analysis of a stiffened cylindrical shell are studied to validate the performance of the proposed method. Results demonstrate that the proposed method converges to the required estimated accuracy while saving considerable computational burdens compared with state-of-the-art approaches.

• Publisher: Springer Berlin Heidelberg,Springer Nature B.V
• Subject: Computational Mathematics and Numerical Analysis / Coupling / Cylindrical shells / Efficiency / Engineering / Engineering Design / Failure analysis / Failure modes / Limit states / Methods / Model accuracy / Monte Carlo simulation / Reliability analysis / Reliability engineering / Research Paper / System reliability / Theoretical and Applied Mechanics