Hybrid Control Strategy With Neural-Network-Assisted Synchronous Rectification for Efficient Wide-Gain <i>CLLC</i> Converter
Zihang Cheng, Liangzong He
Abstract
Synchronous rectification (SR) is an effective method to improve the efficiency of <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">CLLC</i> converters. However, achieving real-time SR is highly challenging due to the complex physical models involved. Moreover, the diverse modulation methods of <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">CLLC</i> further complicate and hinder the implementation of SR. To address those issues, this article proposes a hybrid control strategy, aiming to expand the gain, that utilizes neural-network assistance and online efficiency optimization. By employing a neural-network model instead of complex physical models, the proposed strategy can rapidly and accurately provide initial SR control signals under various operating conditions. Furthermore, the strategy dynamically updates and optimizes these SR control signals based on the online system efficiency optimization to compensate for the error from the proposed hybrid control strategy. Additionally, this approach can adapt to different modulation strategies to meet the wide voltage range requirements of the <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">CLLC</i> converter. The experimental validation conducted on a 400-V/400-V, 800-W experimental prototype affirmed the effectiveness, real-time capability, and versatility of the proposed method.