A Cut-Based BAT-MCS Approach for Binary-State Network Reliability Assessment

TL;DR

This paper introduces cBAT-MCS, an improved hybrid Monte Carlo method for binary-state network reliability that reduces complexity, variance, and runtime, outperforming previous methods in efficiency and accuracy.

Contribution

The paper proposes a novel layer-cut approach for super-vector selection in BAT-MCS, enhancing performance and reducing computational complexity.

Findings

cBAT-MCS outperforms traditional MCS in efficiency and accuracy

Layer-cut super-vector selection reduces runtime and variance

Numerical experiments confirm improved reliability assessment performance

Abstract

The BAT-MCS is an integrated Monte Carlo simulation method (MCS) that combines a binary adaptation tree algorithm (BAT) with a self-regulating simulation mechanism. The BAT algorithm operates deterministically, while the Monte Carlo simulation method is stochastic. By hybridizing these two approaches, BAT-MCS successfully reduces variance, increases efficiency, and improves the quality of its binary-state network reliability. However, it has two notable weaknesses. First, the selection of the supervectors, sub-vectors that form the core of BAT-MCS, is overly simplistic, potentially affecting overall performance. Second, the calculation of the approximate reliability is complicated, which limits its strength in reducing variance. In this study, a new BAT-MCS called cBAT-MCS is proposed to enhance the performance of the BAT-MCS. The approach reduces the complexity of MCS. Selecting the super-vector based on a novel layer-cut approach can reduce both runtime and variance. Extensive numerical experiments on large-scale binary-state network demonstrate that the proposed new cBAT-MCS outperforms traditional MCS and original BAT-MCS approaches in terms of computational efficiency and accuracy.