Resilience assessment and intelligent allocation of emergency repair resources for transmission and distribution networks under typhoon disasters

Abstract

Typhoons can cause correlated failures across transmission corridors, distribution feeders, roads, and communication links. Delayed damage confirmation also makes first-come-first-served repair inefficient. This paper proposes a closed-loop resilience assessment and emergency-resource allocation method for coupled transmission and distribution networks. A parametric typhoon field produces wind and rainfall scenarios. SCADA alarms and unmanned-aerial-vehicle RGB/infrared observations then update component fragility priors through Bayesian evidence fusion. The posterior probabilities drive a dynamic service metric that considers transfer capability, supply continuity, and critical-load priority. A graph-attention model ranks repair, switching, inspection, and mobile-energy-storage actions. A rolling-horizon mixed-integer optimizer enforces crew, road, radiality, and power-flow constraints. Tests use 300 Monte Carlo scenarios on a modified IEEE 24-bus system linked to three IEEE 33-bus feeders and a 42-node road graph. Relative to a two-stage stochastic benchmark, the method raises the resilience index from 0.774 to 0.819 and reduces expected energy not served from 8.67 to 6.96 MWh. It also advances 95% critical-load restoration from 10.0 to 8.3 h and cuts mean solve time from 128.4 to 31.7 s. Ablation results confirm that optical updating and rolling feedback are both essential. The method supports auditable emergency decisions and coordinated closed-loop restoration.