A structural reliability analysis method based on Kriging model with adaptive nested sampling

The Kriging model with adaptive sampling is a popular surrogate model for reliability analysis. It selects experimental points based on specific criteria and replaces the performance function for failure probability estimation once the desired accuracy is achieved. However, for small failure probabilities or computationally intensive performance functions, it often exhibits slow convergence due to high computational costs and sample requirements. This paper proposes a structural reliability analysis method based on Kriging Model with Adaptive Nested Sampling (K-ANS). K-ANS refines sampling boundaries iteratively using conditional probability and selects experimental points within subsets through EFF and U learning functions. The Kriging model updates until convergence, and Monte-Carlo Simulation is used for reliability analysis. Validated through two numerical examples and a cable-stayed bridge, K-ANS significantly reduces performance function evaluations while maintaining accuracy, demonstrating its efficiency and applicability to complex engineering structures. Furthermore, the sampling method is versatile and applicable to other surrogate models.