时域
计算
电子工程
整流器(神经网络)
计算机科学
电气工程
工程类
算法
计算机视觉
人工神经网络
随机神经网络
机器学习
循环神经网络
作者
Haoran Li,Tong Lei,Cungang Hu,Xirui Zhu,Kun Tan,Xi Tang,Xiaoyong Ren,Zhiliang Zhang,Wenping Cao
标识
DOI:10.1109/tpel.2024.3456859
摘要
It is extremely important to use synchronous rectifier (SR) as the diode forward voltage of SiC mosfets may be up to six times higher than Si mosfets, which causes much higher conduction loss. Conventional CLLC SR typically uses detection circuits or builds complex models, but they are susceptible to high dv/dt generated from SiC devices, or have difficulty in implementing SR online due to the complex numerical calculation. A digital real-time computation SR algorithm is proposed for bidirectional SiC CLLC converter. Analytic models in the time domain are constructed to calculate the SR conduction time online. It not only achieves low conduction loss by optimizing the on-time of SR MOSFETs, but also provides high immunity to the high switching frequency noise. A prototype of 300-kHz 6.6-kW SiC bidirectional CLLC charger was built to verify the proposed control. With the proposed SR, the CLLC efficiencies are up to 97.56% and 97.75% at full load in the forward and reverse modes, respectively. Moreover, the CLLC full load efficiencies improves 0.42% and 0.35% over the conventional SR control in the forward and reverse modes. The power density of proposed charger increment is also up to 20% over Wolfspeed charger.
科研通智能强力驱动
Strongly Powered by AbleSci AI