.png){kind=icon}
Return
A single-loop reweighted numerical integration method for failure probability function estimation under parametric probability boxes
.png){kind=icon}
OA
AI
DOI：10.1016/j.ymssp.2026.114964.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Estimating the failure probability function under parametric probability boxes (p-boxes), which capture mixed aleatory and epistemic uncertainties, remains computationally challenging, especially when expensive computer models are involved. To address this challenge, the present paper develops a single-loop reweighted numerical integration method for efficient estimation of the complete failure probability function and its bounds. The generalized failure probability function is reformulated through a reweighting scheme that employs an auxiliary density function. Since the theoretically optimal auxiliary function is intractable, we provide two practical and reliable alternatives. The suggested reweighted probability operation is then combined with Generalized F discrepancy-based point selection strategy to compute the reformulated failure probability in a single-loop manner. This point selection strategy partitions the uncertainty space into Voronoi cells and assigns a representative point with the associated probability to each cell, avoiding brute-force sampling. The framework is fully non-intrusive, as it relies solely on the input p-boxes without relying on the form of the performance function. Finally, four case studies: a mathematical example, a vehicle front axle, a 120-bar truss, and a nonlinear Duffing oscillator, demonstrate that the proposed method achieves high accuracy and efficiency, is applicable to nonlinear and dynamic problems, and integrates seamlessly with black-box simulation codes for complex engineering applications.
Keywords:
Parametric probability box
Hybrid uncertainties
Failure probability function
Numerical integration
Auxiliary density function
Journal
IF:
8.9
Papers:
1.3W
Citations:
6.6W
Organization
Cited Papers
No cited papers available