A cuckoo search algorithm applied to the electric grid interdiction problem

Abstract

ABSTRACT: The Electric Grid Interdiction Problem (EGIP) considers the interaction of a disruptive or malicious agent and the system operator. The disruptive agent pretends to maximize damage to the network; for this he must decide a set of lines to attack in order to maximize load shedding. The independent system operator reacts to such attack by redispatching available generation aiming to minimize load shedding. The interaction of both agents is modeled as a Stackelberg leader-follower game and framed in a bilevel programming structure. Due to its non-convexity, the EGIP has been traditionally approached by means of linearized equivalents of the network. In this paper we used a nonlinear modeling of the network and expressed the EGIP as a mixed integer non-linear programming (MINLP) problem providing more accurate results. The model is solved by means of a cuckoo search algorithm which performance is compared with a hybridized genetic algorithm and a traditional mixed integer linear programming (MILP) approach. The proposed algorithm provides valuable information to the system operator and the system planner regarding the most critical lines. Results show the applicability and robustness of the proposed approach.