Performance Evaluation of an RL-Aided DAB-Based EV Charger Control

Dual Active Bridge (DAB) dc-dc Converter has drawn attention for electric vehicle charging applications due to its ascendancy such as higher power density, bidirectional mode of operation, high efficiency and many more. Different approaches has been taken so far to boost its performance. This study presents performance analysis of a single phase shifted DAB converter to achieve desired output current for electric vehicle (EV) charging applications. The applied DAB converter has a proportional integral (PI) controller unit tuned by reinforcement learning (RL) based Deep Deterministic Policy Gradient (DDPG) algorithm. The results show that, the conventionally tuned PI controller has higher rate of overshoot under short circuit fault condition comparing to the RL-aided PI controller. The method and results are validated through simulations in MATLAB/Simulink.