Model-Free Neural-Network-Based Adaptive Control for Single-Phase Dual-Active-Bridge Converter

• Published in: 2022 IEEE 1st Industrial Electronics Society Annual On-Line Conference (ONCON) pp. 1 - 7
• Main Authors: Iskandarani, Hassan, Kanaan, Hadi Y.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 09.12.2022
• Subjects: Adaptation models; Adaptive control; Bridge circuits; DAB; DC-DC power converters; Mathematical models; neural network; PI control; Software packages; SPS modulation; voltage regulation
• DOI: 10.1109/ONCON56984.2022.10126668

The Dual Active Bridge (DAB) DC-DC converter have several uses in current energy architectures, because of its numerous advantages, it is always possible to find the DAB in micro grids applications, energy storage systems applications, vehicles to grid applications, and a lot more. This wide range of applications subject the DAB to system variations and disturbances on both input side and output side, causing deficient performance of the DAB. This study proposes a model-free adaptive control based on feed-forward neural network, to control the output voltage of the DAB and to maintain it constant under system variation with finite time response. The proposed controller has the same layout as a PI controller. The study is done using MATLAB Simulink, where the system is tested under system variations. A performance test using time domain analysis is done for the proposed controller, a PI controller, and to a combination of an AANN in parallel with a PI controller. The comparison between the three controllers is concluded, and showed the upper hand for the proposed controller.