Rapid resilience assessment method for large-scale road transportation systems based on information value-driven critical node identification

Rapid physical resilience assessment of large-scale road networks is computationally challenging due to the quadratic growth of origin-destination (OD) pairs. To overcome this challenge, a resilience assessment method is proposed based on information value-driven critical node identification. Critical nodes are identified based on information value, which is defined as the loss in global network efficiency upon node failure. The top-ranked nodes are retained to construct a reduced OD matrix, and shortest paths among retained nodes are cached for reduced per-step computation. The method is tested on two large-scale road networks in China under earthquake and flood scenarios. A sensitivity analysis with a 10% step increment from 10% to 100% of nodes shows that 10% retention causes unacceptable error, and over 40% yields marginal gains with steep runtime cost. The 20% retention level provides the best trade-off, maintaining a relative error in the resilience index below 5% and reduces computational time from tens of hours to minutes, while achieving a normalised root-mean-square error of < 0.08 and an R2 > 0.91 for the resilience curve. The information-value metric consistently outperforms degree, betweenness, closeness, and PageRank centrality. The proposed method offers an efficient, scalable pathway for rapid resilience assessment.