Global failure probability function estimation based on an adaptive strategy and combination algorithm. (March 2023)

• Record Type: Journal Article
• Authors: Yuan, Xiukai; Qian, Yugeng; Chen, Jingqiang; Faes, Matthias G.R.; Valdebenito, Marcos A.; Beer, Michael
• Is Part Of: Reliability engineering & system safety. Volume 231(2023)
• Publication Date: 2023-03
• Subjects: Failure probability function -- Importance sampling -- Combination algorithm -- Adaptive strategy; Reliability (Engineering) -- Periodicals; System safety -- Periodicals; Industrial safety -- Periodicals; Fiabilité -- Périodiques; Sécurité des systèmes -- Périodiques; Sécurité du travail -- Périodiques; 620.00452
• DOI: 10.1016/j.ress.2022.108937
• Languages: English
• ISSNs: 0951-8320

Abstract

Abstract: The failure probability function (FPF) expresses the probability of failure as a function of the distribution parameters associated with the random variables of a reliability problem. Knowledge on this FPF is of much relevance for reliability sensitivity analysis and reliability-based design optimisation. However, its calculation is usually a challenging task. Therefore, this paper presents an efficient approach for estimating the FPF based on an adaptive strategy and a combination algorithm. The proposed approach involves three basic elements: (1) a Weighted Importance Sampling approach, which allows determining local FPF estimates; (2) an adaptive strategy for determining at which realisations of the distribution parameters it is necessary to perform local FPF estimation; and (3) an optimal combination algorithm, which allows to aggregate local FPF estimations together to form a global estimate of the FPF. Test and practical examples are presented to demonstrate the efficiency and feasibility of the proposed approach. Highlights: A global approach for failure probability function estimation is proposed. It utilises weighted importance sampling to efficiently obtain local estimators. It includes an adaptive active strategy to guide the local estimations. It adopts an optimal combination algorithm to combine the local FPF estimators.