A Two-Stage Distributionally Robust Restoration Strategy for Typhoon-Affected Distribution Networks Using a Fuzzy Bayesian Network

Extreme typhoons cause correlated distribution-line outages and disrupt access for restoration resources. This paper formulates a two-stage framework linking state-conditioned fragility, fuzzy Bayesian evidence updating, tail-aware scenario reduction, distributionally robust planning, and coordinated restoration. A reproducible physics-informed Monte Carlo experiment evaluates source-to-bus connectivity and heuristic crew staging over 2,000 paired events generated from a common state-conditioned hazard model. Prior- and posterior-informed staging select the same crew bases, and both yield a mean unserved energy of 29.612 MWh. Energy-limited local mobile storage reduces the mean to 28.173 MWh, a reduction of 4.86%. Bayesian updating changes the conditional source-disconnection risks of Buses 17 and 30 from 0.665 to 0.974 and from 0.590 to 0.945, respectively. The paired design holds fault, repair, access, and load realizations identical across policies, thereby isolating the effect of risk updating on deployment decisions. The results demonstrate stable staging under the updated risk distribution and quantify the energy contribution of local mobile storage. Together, the formulation and reproducible mechanism experiment establish a traceable numerical basis for studying the interaction between risk information, resource access, and energy availability in typhoon restoration.