Smart Frequency Control of Cyber-Physical Power System Under False Data Injection Attacks

This study proposes a two-level defense strategy to cope with false data injection (FDI) attacks applied to the control and measurement signals of a two-area load frequency control (LFC) system. In the first level, the combination of a recursive estimator, a residual signal-based detector, and a model-free observer is proposed to detect the attacks and mitigate the attacks' impacts on the system's measurement signals and secondary control commands. Simultaneously with the first level, a novel model-free independent defense strategy, fractional-order brain emotional learning (FOBEL), is developed in the second level to improve the first-level mitigation performance. The FOBEL controller can make robust and fast decisions in systems facing cyber-attacks. Selecting the appropriate values for the FOBEL controller is vital in improving its performance. To this end, a soft actor-critic deep reinforcement learning (SAC-DRL) algorithm is also employed to tune the FOBEL controller's coefficients. The proposed two-level defense scheme efficiency is evaluated via real-time simulations in the OPAL-RT simulator and compared with the various detection and mitigation methods under different scenarios. The experimental results reveal that the presented defense strategy successfully detects FDI attacks and performs significantly better in mitigating them than the other strategies.