AC sparse modeling for false data injection attack on smart gird

Abstract

False data injection attack can bypass the bad data defense process in the power systems and affect the results of state estimation, so that the system security is in danger because of the wrong states obtained in the dispatch center. The traditional false data injection attack is based on the DC power flow model and has many limitations. In this paper, a global attack model and local attack model are proposed separately. The models are not only nonlinear but also sparse. The proposed sparse attack models take the smart grid as the attack object and minimize the change of the tampering meter as the objective function and take the line overload and the node power balance as the constraints. The attack vector obtained by the proposed attack model not only satisfies the node power balance constraints, but also can avoid traditional bad data detection process in power systems. Because the line overload will drive the grid into a dangerous running state that may cause the protection malfunction, a large blackout may be resulted. The proposed nonlinear sparse attack model is comprehensively validated on IEEE standard models.