Time-Variant Reliability Analysis with an Adaptive Single-Loop Kriging Combined with Directional Sampling Method

Abstract

The difficulty in calculating time-variant reliability lies in the large number of performance function calls. This paper proposes an efficient method for time-variant reliability analysis by incorporating the directional sampling method (DS). In this method, within the framework of single-loop active learning Kriging (SL–AK), the burden of fitting the Kriging surrogate model is alleviated by constructing a novel candidate sample pool. While reducing the computational burden, the accuracy of the Kriging surrogate model remains unaffected. Unlike the traditional SL–AK, the proposed method obtains the failure probability through an improved DS method. The improved DS utilizes a bisection method for root-finding, thereby circumventing the impact of interpolation coefficients on the computation results. This method of computing failure probability not only alleviates computational burden but also enhances the precision of the estimated failure probability. Furthermore, this paper introduces global sensitive analysis (GSA) into time-variant reliability analysis, thereby expanding the applicability of GSA. The accuracy and high efficiency of the proposed method are verified through three numerical examples and two engineering examples with implicit performance function.