Reliability Analysis Method for Small Failure Probability Based on Adaptive Kriging and Directional Importance Sampling
The reliability assessment of engineering problems often requires computationally expensive finite element analyses and involves small failure probabilities; therefore, the development of efficient reliability analysis methods is essential. In this paper, a reliability analysis method tailored for small failure probabilities, termed AK-DIS, is proposed by integrating directional importance sampling (DIS) with adaptive Kriging (AK). AK-DIS identifies the intersection points between the directional vectors and the current Kriging-predicted limit state surface as a candidate sample pool, thereby achieving a high computational efficiency. The sample with the greatest predictive uncertainty is then selected from this pool to update the Kriging model. After multiple iterations, the AK-DIS procedure converges, at which point the constructed Kriging surrogate can be employed as a substitute for the original performance function in reliability analysis. AK-DIS was evaluated through six examples, including nonlinear problems, multiple failure domains, high-dimensional problem, and finite element models. The results demonstrate that, while maintaining computational accuracy, AK-DIS achieves a higher efficiency than existing methods.
Authors
- 刘永寿 (ORCID: https://orcid.org/0009-0008-1863-2336)
- Xin Fan (ORCID: https://orcid.org/0009-0000-6118-063X)
- Jupeng Fang
- Huikai Chen
- Zheng Li
Institutions
- Northwestern Polytechnical University (CN)
- Northwest Institute of Mechanical and Electrical Engineering (CN)
Publication Details
- Journal
- International Journal of Computational Methods
- Published
- 2026-11-17
- DOI
- https://doi.org/10.1142/s0219876226500647
- Primary Topic
- Probabilistic and Robust Engineering Design
- Type
- article
- Field-Weighted Citation Impact
- 0.00