Supervised Learning for DC-Link Protection of Dual-Active Bridge Converter Against Cyber-Attacks

This paper proposes a supervised artificial intelligence (SAI)-based control strategy to protect the DC link in a single-phase dual active bridge (SP-DAB) converter from cyber-attacks. The proposed approach involves developing an automated system for training and fine-tuning the SAI model to detect and remove cyber-attacks based on the errors of the converter's DC link voltage and output power. The proposed SAI-based DC-link voltage estimation along with an adjusted Proportional-Integral (PI) controller is incorporated by the DC-link voltage under cyberattack to appropriately modify the dynamic part of the converter's phase shift angle. To further inspect the abilities of the proposed cyberattack strategy, the different forms of malicious data including constant data, triangular wave, and sinusoidal wave are applied. The proposed method is evaluated using simulations in MATLAB/SIMULINK environment to show that the method is effective in detecting and removing cyber-attacks on the DC link of the DAB converter.