Stochastic Sequential Restoration Planning with Risk Awareness and Ferroresonance Avoidance

Large-scale blackouts caused by extreme weather events pose critical challenges to modern power systems, disrupting essential services and causing severe economic and social losses. Conventional restoration strategies typically rely on predesigned backbone networks, assuming intact transmission infrastructure and stable operating conditions. However, storms and other extreme events frequently induce extensive physical damages to power system components, invalidating these assumptions and highlighting the need for more adaptive restoration approaches. This study develops a stochastic sequential optimization framework to model and improve post-storm power system restoration under uncertainty in failure probabilities arising from incomplete information about component health status. Using weather data, the failure probabilities of transmission lines are estimated and incorporated into a markov decision process (MDP) formulation. We consider uncertainty in line failure probabilities but assume that operators can update their estimation based on the latest available information as restoration progresses, which aligns with real-time situational awareness practices. The power grid is represented as a graph, where operators sequentially energize components to reestablish connectivity between substations. The model enforces critical operational and safety constraints, such as generation capacity limits, power flow constraints, and ferroresonance avoidance rules, often neglected in existing studies, to ensure both efficient and secure restoration. The proposed model enhances grid resilience by enabling adaptive restoration planning that dynamically responds to evolving system conditions. To the best of our knowledge, this is the first study to incorporate uncertain failure probability and ferroresonance avoidance explicitly into power system restoration modeling.

Keywords: risk-aware restoration, stochastic sequential optimization, markov decision process, storm-induced failures, resilience optimization.