Identification of key components after unintentional failures for cascading failure protection

Abstract

Cascading failure can aggravate the vulnerability of power grids, which brings attention to cascading failure protection research. Existing works focus on either finding the critical components whose failure can cause large-scale blackouts or methods to mitigate failures after they have happened. However, they are not able to proactively protect against real-world failures, which may not only happen at the critical components. In this paper, we study the problem of finding components that will be impacted the most after unintentional initial failures, which suits the need for practical scenarios. The problem is challenging since approaches like simulating a large number of cascading failures cannot scale and they must be redone when power network parameters change. To tackle the problem, we derive a line importance metric based on all paths and illustrate how it is correlated with highly impacted lines after unintentional failure both intuitively and with an IEEE test case. Further, we design a path sampling algorithm to estimate the metric with provable guarantee and achieve scalability. We evaluate the performance of the proposed method within a protection scenario using various IEEE test cases and demonstrate its superiority against several baseline methods.