.png){kind=icon}
.png){kind=icon}
Return
Structural dominant failure modes analysis method based on limit state sample searching with supervised Laplacian eigenmaps
.png){kind=icon}
PRE
AI
Y
Z
X
M
H
DOI：10.1016/j.strusafe.2026.102743.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
• A probabilistic decoupling framework is proposed for identifying dominant failure modes without explicit modeling of inter-mode correlations, overcoming limitations of traditional failure tree methods. • Supervised Laplacian Eigenmaps (SLE) effectively reduce high-dimensional limit state samples into a low-dimensional mapping space, achieving clear inter-mode separation and robust clustering validation. • Uncertainty-guided active learning with entropy-based sampling iteratively discovers new failure modes, enhancing exploration efficiency in complex structural systems. • The method demonstrates superior computational efficiency across three structural case studies, identifying more failure modes with significantly fewer structural analyses compared to MCS and existing methods. • Applicability to high-dimensional and complex systems is validated through a truss arch bridge with 283 random variables and 137 components, showcasing practical relevance for large-scale engineering structures.
Journal
IF:
6.3
Papers:
1.4K
Citations:
7.0K
Organization
Cited Papers