Efficiency optimization for two-phase interleaved parallel DC-DC converter system

Aiming at the problem that the current-sharing control scheme could not ensure the efficiency of the two-phase interleaved parallel direct current-direct current (DC-DC) converter system to be optimal，resulting in energy waste，an efficiency optimization control strategy based on current distribution was proposed．The support vector regression (SVR) algorithm was used to train the single-phase converter efficiency prediction model to predict the efficiency characteristic curve of each phase converter under arbitrary operating conditions，and then the Q-learning algorithm was adopted to find the best current distribution scheme that could make the system total efficiency optimal．This method overcomes the shortcomings of traditional methods that need to recollect data to fit the efficiency characteristic curve when the operating condition is changed，and avoids tedious calculation of mathematically solving the optimal allocation scheme under complex operating conditions．The model has high applicability and requires less parameters，which can adapt to the dynamic changes of system operating conditions，and give the optimal current distribution scheme without resetting the model．Simulation experiments show that，compared to the current-sharing control，the proposed efficiency optimization control method can significantly improve the system overall efficiency，up to 4.5%. © 2022 Huazhong University of Science and Technology. All rights reserved.