Catastrophic cascading failures in power networks

The high demand of electricity makes power networks more vulnerable under cascading failures. Because of the operational dependencies between nodes, the failure of a small set of nodes can cause a large cascade of failures which results in the breakdown of the network. Thus, it is crucial to study the vulnerability of the power network under the cascading failures.In this paper, we study the cascading critical node (CasCN) problem which asks to find a set of nodes whose failure maximizes the number of failed nodes under the effect of cascading failures. We first show that the problem is NP-hard to approximate within the factor of O(n1−ϵ)O(n1−ϵ). We then design a new metric to evaluate the importance of nodes in the network and use it as the base to design the Fully Adaptive Cascading Potential algorithm. In the case where the network is robust, we propose an alternative algorithm, the Cooperating Attack algorithm, which includes several novel properties to solve the problem. Simulation results demonstrate the efficiency of proposed algorithms and provide more insight into the vulnerability of the power network.