Sequential recovery of power system in cascading failure based on Recovery Probability Matrix

Cascading failure in power system may trigger extensive blackouts with severe societal consequences. This paper proposes a probabilistic recovery framework that centers on a Recovery Probability Matrix (RPM). The Recovery Probability Matrix (RPM) quantifies the extent to which actively restoring failed links either facilitates or inhibits the passive recovery of other links. Building on the RPM, a Joint Centrality (JC) is introduced to capture both the systemic influence and structural dependency of links. Three recovery strategies are developed based on the RPM and JC, including Static Batch Recovery, Static Incremental Recovery, and a Dynamic strategy that adaptively adjusts priorities as the network evolves. This Dynamic strategy forms an adaptive recovery chain, where each decision is guided by the current network state and influences subsequent actions. Simulations on the IEEE 39 Bus Test Case show that the Static Batch strategy achieves rapid recovery but may overuse resources, whereas the Dynamic strategy yields stable and resource-efficient performance under moderate and severe stress levels. All RPM-based strategies consistently outperform baseline methods in recovering links, load, and total system performance under both moderate and severe failure scenarios. • Recovery Probability Matrix (RPM) models link-level recovery influence. • Joint Centrality (JC) measures systemic impact and structural dependency. • Three recovery strategies: static batch, incremental, and dynamic. • Dynamic strategy adapts to topology changes and limits secondary failures. • RPM-based strategies are faster and more efficient than baselines; dynamic is best.