Attack Detection for Load Frequency Control Systems Using Stochastic Unknown Input Estimators

False data injection attacks (FDIAs) against the automatic generation control (AGC) system can lead to unstable or non-optimal operation of the power grid. This paper introduces a method to detect FDIAs targeting the AGC system by developing a stochastic unknown input estimator (SUIE). The SUIE estimates the states of the load-frequency control system, which contains the AGC as a control loop. An increase in the SUIE's residual function (RE) beyond a defined threshold signifies an EDI A. The SUIE can be designed such that it works independently from some or all inputs to the system's state-space model. In addition, the effect of process and measurement noise on the estimated states is minimized through an optimal gain setting technique for the SUIE. Therefore, not only does the SUIE eliminate the need for information about real-time load changes throughout the grid, it also maximizes the state estimation accuracy. The combination of these features distinguishes the proposed method from existing FDIA detection techniques for the AGC system. This paper also develops a number of attack identification SUIEs (AISUIEs) to determine which measurements are compromised by an FDIA, thus facilitating FDIA mitigation strategies. The AISUIEs model FDIAs targeting each AGC measurement by an attack input. These inputs serve as the unknown inputs of different AISUIEs, whose RFs indicate the type of attack. The designed AISUIEs also differentiate between attacks and non-attack abnormalities such as faults. Simulation analysis of a three-area power system corroborates the effectiveness of the proposed method. In addition, the performance of the proposed method is tested using an OPAL real-time simulator, and is com Dared with another technique from the literature.