Revealing temporal features of attacks against smart grid

Protecting smart grid against malicious attacks is a major task for the power and energy community. While different initial failures could trigger cascading effects resulting in same or close final impact, the pattern of intermediate cascading process varies significantly. In this paper we propose an approach to analyze temporal features and predict critical intermediate stages to help prevent a massive cascading failure. Initial victims from most vulnerable set of nodes are tested in a topological cascading model under different fault tolerances, and the results reveal that they are usually followed by a dramatic increase of failed components at some critical point. By analyzing the processes of failure propagation, we identify important temporal features of cascading failure and predict critical moments to allow quick and proper response at an early stage. This work provides informative decision support for defense against large blackouts caused either by random contingencies or attack schemes.