Fast Searching Strategy for Critical Cascading Paths Toward Blackouts

Abstract

The analysis of severe cascading blackouts is an essential issue in power grid planning and operation. As there are a tremendous number of possible cascade paths in a large power system, it is very challenging to determine the critical cascading paths that may have catastrophic consequences. This paper proposes a complex network theory based methodology to determine such critical cascading paths with high efficiency. To this end, we construct an improved interaction graph to cope with the situation where the $(N-1)$ criterion is not satisfied during the follow-up cascade propagation after an initial fault. Subsequently, based on the graph, we derive a modified PageRank model to fast rank the influence of individual transmission lines on blackout risks. Further, leveraging the results of influence ranking, we devise an efficient strategy for searching critical cascading paths. Then, we derive an unbiased probability estimation method for individual cascading paths and the blackout. Simulations carried out on the IEEE 39-bus system, the IEEE 118-bus system, and a real-world 1122-bus system in China show that our method can enhance the searching efficiency by up to three orders of magnitude compared with standard Monte Carlo approaches, demonstrating its potential for cascading blackout analysis in large-scale power systems. Results also verify that the proposed probability estimation method is unbiased, and it can provide an efficient way for probability estimation of the blackouts with very low probability but high losses.