Answers are grounded in this paper and cite their sources.
Continuous box girder bridges have been widely used in metropolises as ramps for roadway connections. However, multiple overturning accidents have occurred in recent years, raising significant safety concerns. Evaluation of the anti-overturning reliability of such bridges is critical for supporting decision making in bridge operation and management. Dynamic finite element analysis (DFEA) is regarded as a high-fidelity tool for comprehensively simulating bridge performance. Nevertheless, its high computational cost and discrete output characteristics pose substantial challenges to the application of classical reliability analysis methods. To address these challenges, this paper develops a parallel active-learning approach for surrogate modeling-based reliability analysis, called Par-PSVM-RA. Specifically, several key factors critical to parallel active learning are discussed, and a filtering-then-clustering strategy is employed for parallel active learning. The performance of the proposed method is validated using two numerical examples and an engineering case study of a practical bridge accident. The influence of the parallel sampling number is further investigated, revealing a trade-off between convergence speed and total computational cost. By integrating Par-PSVM-RA within a parametric DFEA framework, reliability indices under different traffic scenarios are evaluated to provide guidance for practical engineering decision making. Several factors influencing the anti-overturning capacity of bridges are discussed in detail, offering insights into future engineering applications and risk management for bridge infrastructure.
No takes yet. Share an insight, caveat, or question.
Song et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: