基于模糊神经网络的PEMFC输出电压自抗扰控制策略  

PEMFC output voltage active disturbance rejection control strategy based on fuzzy neural network

作　　者：杨旭红 于嘉炜 张苏捷 钱峰伟 Yang Xuhong;Yu Jiawei;Zhang Sujie;Qian Fengwei(School of Automation Engineering,Shanghai Electric Power University,Shanghai 200090,China;Shanghai Electric High Voltage Industrial Co.,Ltd.,Shanghai 200333,China;Shanghai Solar Energy Engineering Technology Research Center,Shanghai 200241,China)

机构地区：[1]上海电力大学自动化工程学院,上海200090 [2]上海电力高压实业有限公司,上海200333 [3]上海太阳能工程技术研究中心,上海200241

出　　处：《电子测量技术》2025年第4期62-70,共9页Electronic Measurement Technology

基　　金：国家自然科学基金(51777120);上海市2021年度“科技创新行动计划”科技支撑碳达峰碳中和专项(第一批)(21DZ1207502)项目资助。

摘　　要：质子交换膜燃料电池存在输出电压不稳定,发电效率低下等问题,需要使用Boost电路进行升压,以此确保电压质量,满足系统需求。根据PEMFC的输出特性,在Matlab/Simulink平台搭建PEMFC以及Boost电路的数学模型,考虑线性自抗扰控制策略对扰动具有优异的动态响应速度,提出一种基于模糊神经网络的线性自抗扰控制策略,用于Boost电路的电压环控制,依靠模糊神经网络对线性自抗扰控制器中的关键参数进行整定,以实现控制器的实时优化。通过仿真分析对比不同工况下,FNN-LADRC控制策略与LADRC控制策略下输出电压的性能差异,结果显示,在无扰动情况FNN-LADRC控制策略下的调节时间为5 ms,LADRC控制策略下的调节时间为40 ms,在扰动情况时FNN-LADRC控制策略调节时间更快,抗干扰能力更强。结合绝对误差积分IAE指标和时间乘绝对误差积分指标ITAE指标进行系统整体性分析,验证了所提控制策略的有效性与优越性。Proton exchange membrane fuel cell(PEMFC)has problems such as unstable output voltage and low power generation efficiency,so Boost circuit is needed to ensure voltage quality and meet system requirements.According to the output characteristics of PEMFC,the mathematical models of PEMFC and Boost circuit are built on the Matlab/Simulink platform.Considering that linear active disturbance rejection control(LADRC)has excellent dynamic response speed to disturbance,a fuzzy neural network-linear active disturbance rejection control(FNN-LADRC)is proposed for voltage loop control of Boost circuits.The key parameters of the linear active disturbance rejection controller are tuned by FNN-LADRC to realize real-time optimization of the controller.The simulation analysis compares the performance difference of output voltage between the FNN-LADRC control strategy and the LADRC control strategy under different working conditions.The results show that the regulation time under the FNN-LADRC control strategy is 5 ms and the regulation time under the LADRC control strategy is 40 ms without disturbance.The FNN-LADRC control strategy has faster adjustment time and stronger anti-interference ability under disturbed conditions.Combined with IAE index and ITAE index,the effectiveness and superiority of the proposed control strategy are verified.

关 键 词：质子交换膜燃料电池 模糊神经网络 线性自抗扰 输出电压 

分 类 号：TM46[电气工程—电器] TN7[电子电信—电路与系统]