Resilience assessment of power distribution systems based on structure varied dynamic Bayesian network under rainstorm disasters

Resilience assessment is crucial in assessing the power supply capacity and guiding the planning and operation of the distribution systems. In recent years, frequent rainstorm-induced urban waterlogging poses a severe threat to power distribution systems, resulting in widespread and prolonged outages. This paper proposes a model-based resilience assessment framework to reveal the mechanistic impacts of such catastrophes on the structural disintegration of system topologies and their cascade effects on operational strategies. In the framework, the rainstorm and two-dimensional hydrodynamics model under spatiotemporal evolution are formulated combining the digital elevation model, and the rainstorm and waterlogging impact on critical infrastructures (CIs) is assessed as prior failure possibility. A structure varied dynamic Bayesian network is proposed combining with the prior failure possibility of CIs to achieve a posterior failure possibility considering the evolution of the system topology. The vulnerable nodes are identified and the system resilience is evaluated according to the redistribution of the expected optimal power flow under the cascade spread of disasters. Finally, the proposed framework is applied to the IEEE 33-node system. It is prove that the system resilience is overestimated by 15.96%, only considering the prior failure probability of the CIs. By relocating the batteries to the three most vulnerable nodes that identified based on the SVDBN approach, the system losses can be reduced by 5.6%.