Rare-event splitting simulation for analysis of power system blackouts

Severe blackouts due to cascading failures in electric grid are rare but catastrophic. The analysis of such blackouts may include evaluation of rare-event probabilities which is difficult to estimate. In this paper, we present an effective rare-event simulation technique to estimate the rare-event probability. We apply our technique to an IEEE-bus electric network and show that our technique can dramatically improve the simulation efficiency. We also demonstrate that we can effectively detect the most vulnerable link in the electric grid, which has the highest probability of leading to a blackout event.