Power system functionality is determined by the evaluation of the power system state estimation (PSSE). Thus, corruption of the PSSE may lead to severe consequences, such as financial losses, substantial maintenance damage, and disruptions in the electricity distribution. Classical bad data detection (BDD) methods, developed to ensure PSSE reliability, are unable to detect well-designed attacks, named unobservable false data injection (FDI) attacks. In this paper we develop novel sparse methods for the detection of unobservable FDI attacks and for the identification of the attacked buses' locations. Furthermore, the proposed methods can be used for estimation of the attack's extent. Hence, it also enables evaluation of the PSSE by removing the estimated attack from the measurements. The proposed methods are based on formulating structural, sparse constraints on both the attack and the load demand. First, we exploit these constraints in order to compose an appropriate model selection problem. Then, we develop the associated generalized information criterion (GIC) for this problem, which is widely employed in signal processing. For large networks, the GIC selection method is time consuming with computational complexity that grows exponentially with the network size. Thus, we develop two low-complexity methods: 1) an orthogonal matching pursuit (OMP)-based method that relies on the proposed structural and sparse constraints; and 2) a new method that exploits the graph Markovian property of order two of the graph representation of the electricity network, i.e. the second-neighbor relationship between the power system buses. The performance of the proposed methods is evaluated on IEEE standard test systems.
|State||Published - 19 Mar 2020|