Imperfect Nonlinear False Data Injection Attack Against Largest Normalized Residual Test

Abstract

False data injection attack is one of the most important cyber-threats to the emerging smart grid. The original perfect attack was developed for DC power flow model which is not used much in practice. A nonlinear version of the attack is also proposed and it showed that it requires real-time errorfree true values of the state variables. In reality, the true values may be difficult to achieve that leads to an imperfect attack. Even though it is known that the imperfect attack will change the norm of the residues, the probability of being detected is not analyzed in detail. In this paper, it is shown that the nonlinear imperfect attack is strong enough to bypass the bad data detection and identification methods with high probability. The efficacy of the attack is shown through simulations on IEEE 30 and 118-bus test systems.