摘要
This paper introduces layer-cut BAT-MCS, an enhanced algorithm for binary-state network reliability assessment. The original BAT-MCS integrates the deterministic Binary Addition Tree (BAT) algorithm with stochastic Monte Carlo simulation (MCS) in terms of the novel supervectors, creating a self-regulating mechanism that reduces variance and improves efficiency. Despite its advantages, BAT-MCS exhibits limitations in supervector selection methodology and computational complexity of approximate reliability calculations. The proposed layer-cut BAT-MCS addresses these weaknesses through a novel layer-cut approach for supervector selection that significantly outperforms traditional min-cut methods. This innovation simplifies MCS complexity while maintaining comprehensive network analysis capabilities. Extensive numerical experiments conducted on small and medium-sized binary-state networks demonstrate that layer-cut BAT-MCS achieves superior computational efficiency and accuracy compared to both traditional MCS and the original BAT-MCS implementations. The results indicate that the layer-cut technique provides a more efficient network decomposition strategy, substantially reducing both runtime and variance. These improvements make layer-cut BAT-MCS particularly valuable for reliability assessment of small-scale or sparse network systems where computational resources are limited and high accuracy is required. © 2025