Neural network-aided 4-DF global efficiency optimal control for the DAB converter based on the comprehensive loss model

Abstract

—Intending to enhance the efficiency of dual-active-bridge (DAB) converters, this paper presents a four degree-of-freedom (DF) efficiency optimization method based on the comprehensive loss model. Besides the traditional triple-phase-shift modulation, the switching frequency is added as another control DF to minimize the core loss of the high-frequency transformer (HFT). Considering the nonlinear characteristics of the switching devices and the HFT, the comprehensive loss model is established first to reflect the converter's overall loss directly. Then the algorithm-based efficiency optimization solution is proposed to investigate the 4-DF modulation variables, which are subject to the minimal overall loss of the DAB converter in a given operating condition. A neural network (NN) is then adopted to model the mapping relationship between the ideal 4-DF variables and the working conditions. On this basis, a closed-loop controller that combines the traditional PI regulator and the NN module is suggested to realize both the output target control and the efficiency optimization. Finally, the performance of the proposed control strategy is fully verified through a 1.2 kW experimental prototype. The experimental results show that the converter's efficiency is significantly improved with the NN-aided 4-DF control strategy.