.png){kind=icon}
.png){kind=icon}
Return
An efficient importance sampling method for time-variant reliability analysis
.png){kind=icon}
PRE
AI
DOI：10.1016/j.probengmech.2026.104037.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Time-variant reliability problems commonly arise in engineering systems owing to material degradation and stochastic loading modeled as random processes. Evaluating the probability of failure of a structure or system over its entire service period is therefore of substantial practical importance. This paper develops an efficient importance sampling method based on the First-Order Time-variant Reliability Expansion (FOTRE), referred to as FOTRE-based importance sampling (IS). First, the time-variant failure event is represented as a series system failure event after time discretization. The instantaneous most probable points and reliability indices are then efficiently approximated using FOTRE and employed to construct a weighted mixture importance sampling probability density function. Second, under the proposed importance sampling distribution, a Gaussian process model of the time-dependent response is constructed using the first-order response expression provided by FOTRE. Third, an active learning strategy is developed to identify the critical time locations of each importance sample trajectory and evaluate its failure indicator with only a limited number of performance function evaluations. The proposed framework therefore reduces both the sample size required to achieve a prescribed statistical accuracy and the number of function calls required for each trajectory. Numerical examples demonstrate the accuracy and computational efficiency of the proposed method.
Keywords:
Time-variant reliability
FOTRE method
Importance sampling
Active learning strategy
Journal
IF:
3.5
Papers:
1.7K
Citations:
4.1K