A Deep Learning Based Cyber Attack Detection and Mitigation Scheme in Synchronous Buck Converter

Power electronic converters (PEC) play a vital role in the recent advancements of electric vehicles (EV). Developing an effective converter architecture resilient to cyber-attacks is a need of the hour. In this article, deep learning (DL) controlled synchronous buck converter is designed to make the system perform efficiently during parametric variations and cyber-attacks like false data injection (FDI) attacks. The DL methodology is used to design the control algorithm; it consists of a combination of two neural networks, one to produce duty pulses to the respective switches and the other to detect and mitigate FDI attacks. Real-time hardware implementation is performed to validate the proposed methodology, and its results are explored.