质子交换膜燃料电池故障诊断的模式识别方法  

Pattern Recognition Method for Proton Exchange Membrane Fuel Cell Fault Diagnosis

作　　者：周苏[1,2] 韩秋玲[1] 胡哲[1] 杨铠[1] 

机构地区：[1]同济大学汽车学院,上海201804 [2]同济大学中德学院,上海201804

出　　处：《同济大学学报（自然科学版）》2017年第3期408-412,共5页Journal of Tongji University:Natural Science

基　　金：高等学校创新引智计划(B08019);上海市重点学科项目(B303)

摘　　要：针对质子交换膜燃料电池(proton exchange membrane fuel cell,PEMFC)故障诊断,提出了一种基于电化学交流阻抗谱(EIS)的组合模式识别方法,即采用典型频率点的EIS拟合,快速获取EIS并达到抑制低频点"飘散"的效果,以EIS与实轴的两个交点(高频交点和低频交点)作为故障诊断的特征向量值,采用模糊C均值聚类和支持向量机算法分别对样本特征向量、待测特征向量进行聚类和分类.采用已发表文献中的实验数据构造了故障模型并进行了仿真分析,结果表明提出的方法快速、有效.Based on electrochemical impedance spectroscopy （EIS） ,a combined pattern recognition method is proposed for proton exchange membrane fuel cell （PEMFC） fault diagnosis. By fitting typical frequency points, quickly drawing EIS for suppressing low-frequency points ＂drift＂, taking two intersections with the real axis of the EIS （high frequency intersection and low frequency intersection）as fault diagnosis eigenvector values, a combined algorithm of Fuzzy C-means Algorithm and Support Vector Machine is used to cluster sample eigenvectors and classify under test eigenvectors. Simulation is developed in MATLAB by using experimental data from the published papers, results show the proposed method is fast and effective.

关 键 词：质子交换膜燃料电池 电化学交流阻抗谱 故障诊断 模式识别 

分 类 号：TK91[动力工程及工程热物理]