引用本文: | 罗 朋,樊涵宇,梁剑鑫,姜淏予,刘洺辛.基于神经网络拟合显式MPC的高增益直流变换器[J].电力系统保护与控制,2023,51(20):47-61.[点击复制] |
LUO Peng,FAN Hanyu,LIANG Jianxin,JIANG Haoyu,LIU Mingxin.High gain DC converter based on a neural network fitting explicit MPC[J].Power System Protection and Control,2023,51(20):47-61[点击复制] |
|
摘要: |
针对燃料电池、风力发电等新型能源系统输出电压较低的问题,提出一种具有拟合显式模型预测控制(model predictive control, MPC)的高增益、高效率直流变换器。该变换器具有非隔离型三绕组耦合电感基本单元结构,通过改变匝数比,在合适占空比下实现高增益,同时也降低了开关器件应力。此外,所提变换器采用无源钳位电路,回收漏感能量,抑制了开关管的电压尖峰。为了提高所提变换器的动态性能及抗扰能力,利用神经网络离线拟合显式MPC控制规律的策略,提高了输出电压跟踪精度,减小了输入电压变化和负载变化带来的扰动,具有良好的动态响应。最后在理论分析的基础上,制作出了一台输入10~12 V、输出100 V/100 W的实验样机,实验结果验证了所提变换器的有效性。 |
关键词: 高增益直流变换器 三绕组耦合电感 神经网络 显式模型预测控制 |
DOI:10.19783/j.cnki.pspc.221830 |
投稿时间:2022-11-17修订日期:2023-02-28 |
基金项目:国家自然科学基金项目资助(62272109,62171143) |
|
High gain DC converter based on a neural network fitting explicit MPC |
LUO Peng,FAN Hanyu,LIANG Jianxin,JIANG Haoyu,LIU Mingxin |
(College of Electronic and Information Engineering, Guangdong Ocean University, Zhanjiang 524088, China) |
Abstract: |
Given the low output voltage of new energy systems, such as fuel cells or wind power, this paper proposes a high gain and high efficiency DC converter with fitting explicit model predictive control (MPC). The converter has the basic unit structure of non-isolated three-winding coupled inductor. By changing the turns ratio, this not only achieves higher voltage gain with the appropriate duty cycle, but also reduces the switching device stress. In addition, a passive clamp circuit is adopted to recycle the leakage inductance energy and suppress the voltage spike of the switch. To improve the dynamic performance and immunity of the proposed converter, the strategy of offline fitting of explicit MPC control law of a neural network is used to improve the output voltage tracking accuracy, and reduce the disturbance caused by variational input voltage and load, achieving good dynamic response. Finally, based on theoretical analysis, a prototype with input of 10~12 V and output of 100 V/100 W is built to verify the effectiveness of the proposed converters. |
Key words: high gain DC converter three-winding coupled inductor neural network explicit MPC |