基于二阶RQ-RLC模型的质子交换膜燃料电池水管理故障诊断  

Second-order RQ-RLC Model-based Fault Diagnosis for Water Management in Proton Exchange Membrane Fuel Cells

作　　者：刘相万 杨扬[1] 朱文超 谭金婷 谢长君[1] LIU Xiangwan;YANG Yang;ZHU Wenchao;TAN Jinting;XIE Changjun(School of Automation,Wuhan University of Technology,Wuhan 430070,Hubei Province,China;Hubei Key Laboratory of Advanced Technology for Automotive Components(Wuhan University of Technology),Wuhan 430070,Hubei Province,China;State Key Laboratory of Advanced Technology for Materials Synthesis and Processing(Wuhan University of Technology),Wuhan 430070,Hubei Province,China)

机构地区：[1]武汉理工大学自动化学院,湖北省武汉市430070 [2]现代汽车零部件技术湖北省重点实验室(武汉理工大学),湖北省武汉市430070 [3]材料复合新技术国家重点实验室(武汉理工大学),湖北省武汉市430070

出　　处：《中国电机工程学报》2022年第21期7893-7904,共12页Proceedings of the CSEE

基　　金：国家重点研发计划项目(2020YFB1506802);广东省重点领域研发计划项目(2020B0909040004)。

摘　　要：水淹和膜干故障严重影响质子交换膜燃料电池(proton exchange membrane fuel cell,PEMFC)的工作性能及使用寿命。为充分刻画高频及低频段电化学反应信息,该文建立宽频PEMFC电化学阻抗谱,提出基于二阶RQ-RLC等效电路模型的水管理故障诊断方法。首先,搭建燃料电池测试台架,进行水管理故障模拟实验,测试得到对应的电化学阻抗谱,辨识二阶RQ-RLC等效电路模型的参数,并获得八维水管理故障数据集。然后,运用线性判别分析方法对高维水管理故障数据集降维得到故障特征样本集,并选取4个模型关键参数作为故障诊断特征量。最后,提出自适应差分进化优化支持向量机算法对故障特征样本集进行分类,在50组、130组和210组样本下测试集分类准确率分别100%、97.44%和95.24%,结果表明所提方法能准确诊断出燃料电池所处的水管理故障类型。Flooding and drying failure seriously affect the working performance and service life of PEMFC. In order to fully characterize the high and low frequency electrochemical reaction information, wide-band PEMFC electrochemical impedance spectroscopy was established in this paper, and a water management fault diagnosis method based on secondorder RQ-RLC equivalent circuit model was proposed. First, a fuel cell test bench was built and water management fault simulation experiment was conducted to obtain the corresponding electrochemical impedance spectrum, identify the parameters of the second-order RQ-RLC equivalent circuit model, and obtain an eight-dimensional water management fault data set. Then, the linear discriminant analysis method was used to reduce the dimension of the high dimensional water management fault data set to obtain the fault feature sample set, and four key parameters of the model were selected as the fault diagnosis feature parameters. Finally, the adaptive differential evolution optimization support vector machine algorithm was proposed to classify fault feature sample sets,the test set classification accuracies of 100%, 97.44% and 95.24% at 50, 130 and 210 sets of samples, respectively,showing that the proposed method can accurately diagnose the type of water management fault the fuel cell is subjected to.

关 键 词：质子交换膜燃料电池 二阶RQ-RLC等效电路模型 自适应差分进化优化支持向量机 水管理 故障诊断 

分 类 号：TK91[动力工程及工程热物理]