Assessment of power system vulnerability using metaheuristic techniques

Abstract

ABSTRACT: This paper presents a comparison of different metaheuristic techniques applied to the assessment of power systems vulnerability to intentional attacks, also kwon as the electric grid interdiction problem. This problem is described through a bilevel formulation and comprises the interaction between a disruptive agent (attacker) and the power system operator (defender). The attacker is positioned in the upper level optimization problem and aims at finding the set of devices (lines, transformers and generators) that, once simultaneously attacked, would maximize the system load shedding. This problem is constrained by a limit on destructive resources and the response of the power system operator, located in the lower level optimization problem that reacts to the attack by modifying the generation dispatch aiming at minimizing the load shedding. The interdiction problem described in this paper is nonlinear and nonconvex; therefore, four different metaheuristic techniques are implemented and compared for its solution: Genetic Algorithm, GRASP, Iterated Local Search and Tabu Search. Results show that the Iterated Local Search adapts better to this problem obtaining the best rate between quality of solutions and computation time.