Skip to main navigation Skip to search Skip to main content

Adaptive surrogate-assisted sampling pool reduction strategy for low failure probability estimation

  • Hong Zhang
  • , Lukai Song*
  • , Xueqin Li
  • , Yatsze Choy
  • , Hongxin Liu
  • , Fei Tao
  • *Corresponding author for this work

Research output: Journal Publications and ReviewsRGC 21 - Publication in refereed journalpeer-review

Abstract

To improve the efficiency and accuracy of rare-event reliability analysis of complex structures, an advanced adaptive kriging-based candidate sample reduction (AK-CSR) method is proposed by integrating the CSR strategy and the advanced AK method. Through the improved first-order reliability method and the updated kriging model (KM), the accurate most probable failure point can be obtained with KM updating. By domain constraint and distance constraint functions, the CSR strategy can ceaselessly find desired samples to update the KM. The proposed method was verified using three numerical examples and two engineering examples. The results demonstrated that the AK-CSR method can be used to perform rare-event reliability analysis of complex structures and improve computational efficiency while maintaining good accuracy. Moreover, this study offers a useful insight into reliability-based design optimisation of complex structures and enriches the field of structural reliability theory. © 2025 ICE Publishing. All rights reserved.
Original languageEnglish
JournalProceedings of the Institution of Civil Engineers: Transport
DOIs
Publication statusOnline published - 30 Mar 2025

Funding

This work was supported by the National Natural Science Foundation of China (grant nos 52275471 and 52105136) and the Hong Kong Scholar Programme (grant no. XJ2022013). The authors would like to thank these organisations.

Research Keywords

  • active learning
  • importance sampling
  • kriging model
  • mathematical modelling
  • numerical modelling
  • rare-event reliability analysis
  • sample reduction
  • uncertainty, reliability & risk

Fingerprint

Dive into the research topics of 'Adaptive surrogate-assisted sampling pool reduction strategy for low failure probability estimation'. Together they form a unique fingerprint.

Cite this