A robust optimization approach for protecting power systems against cascading blackouts

Abstract

Due to uncertainties and the complicated intrinsic dynamics of power systems, it is difficult to predict the cascading failure paths once the cascades occur. This makes it challenging to achieve the effective power system protection against cascading blackouts. By incorporating uncertainties and stochastic factors of the cascades, a Markov chain model is developed in this paper to predict the cascading failure paths of power systems. The transition matrix of Markov chain is dependent on the probability of branch outage caused by overloads or stochastic factors. Moreover, a robust optimization formulation is proposed to prevent the cascading blackouts by optimal load shedding and generation control for multiple cascading failure paths with relatively high probabilities. Since each state on the cascading failure paths can be described by one convex set, the proposed robust optimization problem is equivalent to the best approximation problem in Euclidean space. Thus, an efficient numerical solver based on Dykstra’s algorithm is employed to deal with the robust optimization problem. In theory, we provide a lower bound for the probability of preventing the cascading blackouts of power systems. Finally, the proposed approach for power system protection is verified by a case study of IEEE 118 bus system.