Fast risk assessment of distribution grid with iterative inference on probabilistic graph

Extreme weather has induced damages to distribution grids globally. To estimate disaster influences, a fast risk assessment method of the distribution grid is proposed in this work, which is based on iterative inference on the probabilistic graphical model of the outage events. The correlations of the branch failures and load outages are modeled as a probabilistic graph, where each vertex binary states as energized and outage status of buses and edges are weighted by failure probability of branches between corresponding buses. Then, outage probability equations are formulated for all vertexes, which can be solved efficiently by the iterative method to enable accurate and efficient inferences on the graph. Finally, considering the cumulative failure probabilities of branches during extreme weather, the fast outage risk assessment is carried out to predict system losses and vulnerabilities. Moreover, the sensitivity of the branch failure probability regarding system outage risk is derived analytically, which helps to identify critical components for risk leveraging. Case studies are carried out on various distribution grids with different scales, which validate the efficacy of the proposed method.