Real-Time Estimation in Cyber Attack Detection and Mitigation Framework for DC Microgrids

DC microgrids (DCMGs) recently have a wide range of applicability in modern power systems due to their compatibility with advanced distributed control strategies. However, the distributed control schemes bring significant challenges to the system operation due to their high dependency on data transmission through communication networks, making them vulnerable to cyber-attacks. Because of their wide range of applications, cyber-attack detection in DCMGs has recently gained considerable attention. In most machine learning (ML)-based cyber-attack detection algorithms, accurate real-time output estimation with a low computation burden is critical. In this paper, an Adaptive Neuro-Fuzzy Inference Systems (ANFIS)-based real-time estimator is proposed for DCMGs. The proposed technique is used in a cyber-attack detection and mitigation framework to estimate output voltage and current for DC/DC converters with acceptable accuracy and computational cost. Cyber-attack detection has been achieved through residual analysis, comparing the estimated and measured data. Finally, the effectiveness of the proposed ANFIS-based real-time estimator for the ML-based cyberattack detection frameworks is validated in a MATLAB/Simulink testbed and compared to other supervised ML algorithms.