Fast assessment method of power outage risks caused by extreme cold disasters based on probabilistic graph model

Abstract

Abstract The extremely cold weather may cause cascading failures of power grid components, resulting in large-scale outages and economic losses. Thus, to make better anti-disaster preparations, it is necessary to assess the system-level risk of the power grid efficiently with extreme weather predictions. This paper proposes a risk assessment method for power grids based on the probabilistic graphical model (PGM). First, the time-varying failure probabilities of power lines are estimated with consideration of the cascading effects of the direct impact of disasters and power flow transfer. Then, load supply reliabilities are fast inferred by solving the probability graph model of power component failure propagations. Finally, the overall outage risk is evaluated by considering the capacity and importance of different loads, which helps to improve anti-disaster decisions. Through theoretical analysis and case studies, the efficacy of the proposed method is verified.