Scholarlink

<p>In the application of wireless power transfer (WPT) technology, the transmission efficiency of the system is severely limited by the impedance mismatch caused by changes in transmission distance and load fluctuations. The traditional impedance matching method has a slow response and limited adaptability when dealing with complex dynamic conditions. To address this issue, an impedance adaptive matching method for WPT systems based on a lightweight customized neural network prediction algorithm was proposed in this paper. Specifically, a fully connected neural network with a structure of 5-12-8-4-46 was first constructed. Then, the voltage of the system, current signals, and current duty cycle D of the DCDC circuit were used as the input of the model to infer the optimal duty cycle of the DCDC circuit of the system. Finally, the impedance matching of the system was achieved by changing the duty cycle of the DCDC circuit. Simulation and experimental results demonstrate that the proposed method outperforms both the conventional perturb and observe (P&amp;O) and the variable step size perturb and observe (PSS-P&amp;O) approaches. It exhibits a faster dynamic response and improved output smoothness under abrupt changes in load and mutual inductance. These enhancements significantly improve the performance and robustness of the WPT system in dynamic operating conditions.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Impedance matching in WPT systems based on custom neural network parameter prediction

  • 期刊论文

  • 发表日期：

    2026年5月4日

查看全文
  • Jin Chang,
  • Yingzhou Guo,
  • Junhao Wu,
  • Haoyang Wang,
  • Cancan Rong,
  • Chenyang Xia

摘要

In the application of wireless power transfer (WPT) technology, the transmission efficiency of the system is severely limited by the impedance mismatch caused by changes in transmission distance and load fluctuations. The traditional impedance matching method has a slow response and limited adaptability when dealing with complex dynamic conditions. To address this issue, an impedance adaptive matching method for WPT systems based on a lightweight customized neural network prediction algorithm was proposed in this paper. Specifically, a fully connected neural network with a structure of 5-12-8-4-46 was first constructed. Then, the voltage of the system, current signals, and current duty cycle D of the DCDC circuit were used as the input of the model to infer the optimal duty cycle of the DCDC circuit of the system. Finally, the impedance matching of the system was achieved by changing the duty cycle of the DCDC circuit. Simulation and experimental results demonstrate that the proposed method outperforms both the conventional perturb and observe (P&O) and the variable step size perturb and observe (PSS-P&O) approaches. It exhibits a faster dynamic response and improved output smoothness under abrupt changes in load and mutual inductance. These enhancements significantly improve the performance and robustness of the WPT system in dynamic operating conditions.