Data-Driven Automatic Generation Control of Interconnected Power Grids subject to Deception Attacks

In this paper, a Data-Driven Adaptive Control (DDAC) technique is proposed for the Automatic Generation Control (AGC) problem of an interconnected power grid subject to Deception Attack (DA). The emergence of the Internet of Things (IoT) and the advancement of communication technologies have provided an opportunity for power system operators and designers to compensate for the lack of an appropriate model using a huge amount of data. However, they have also caused security challenges in the grid due to malicious attackers. This paper focuses on the attack to the control network which carries the AGC signals between the secondary and local primary frequency controllers. Intentional modifications of AGC signals during an attack may result in frequency instability because of saturation in governor signals. To counteract such an attack, a DDAC is suggested for a multi-area power system in which, the system model is dynamically updated using real-time input and output signals. The model includes the attacker's behavior, thus empowering the control system to act against it. The stability of the proposed controller is proved using the Lyapunov stability theory when the DA causes input saturation. Simulation results show that it can successfully tolerate a class of deception attacks and keep the multi-area power grid stable.