부산대학교 도서관

In recent times, the severity of cascading outages leading to power system blackouts has increased due to unexpected extreme weather conditions and cyber-attacks. In this regard, a risk-based assessment is of paramount importance to identify the vulnerable links in power systems. However, the identification of vulnerable links is an NP-hard problem, which becomes more complex in the presence of uncertain renewable power generations and load demand. To this end, this paper proposes a novel and scalable probabilistic risk assessment method for uncertainty quantification of cascading outages. Here, we propose a novel quantile lite-polynomial chaos expansion (Q-Lite-PCE) to predict multiple quantile values of power flows for each line, and then construct conditional density and distribution functions. These distribution functions are then used to compute overload probabilities for all line power flows triggered by N1 contingency. Numerical studies of the proposed method on an IEEE 30-bus test system reveal that the proposed method can evaluate the probability of cascading outages accurately in the presence of uncertainly correlated nodal power injections. Specifically, it can achieve a mean %$L_{1} -$relative error of around 1. 5% in power flow predictions and of less than 0.5% in overload probability estimations.