电工技术学报  2019, Vol. 34 Issue (18): 3949-3960    DOI: 10.19595/j.cnki.1000-6753.tces.181051
电化学储能 |
基于在线序列超限学习机和主成分分析的蒸汽冷却型燃料电池系统快速故障诊断方法
刘嘉蔚, 李奇, 陈维荣, 余嘉熹, 燕雨
西南交通大学电气工程学院 成都 611756
Fast Fault Diagnosis Method of Evaporatively Cooled Fuel Cell System Based on Online Sequential Extreme Learning Machine and Principal Component Analysis
Liu Jiawei, Li Qi, Chen Weirong, Yu Jiaxi, Yan Yu
School of Electrical Engineering Southwest Jiaotong University Chengdu 611756 China
全文: PDF (30066 KB)  
输出: BibTeX | EndNote (RIS)      
摘要 为解决蒸汽冷却型燃料电池系统的故障诊断问题,该文提出基于在线序列超限学习机和主成分分析的蒸汽冷却型燃料电池系统快速故障诊断新方法。新方法采用主成分分析过滤冗余信息,得到能反映蒸汽冷却型燃料电池系统状态的故障特征向量;使用在线序列超限学习机对故障特征向量进行分类,能有效提高模型诊断正确率并降低运算时间。实例分析表明:新方法可快速识别膜干故障、氢气泄漏故障和正常状态共三种健康状态。算法的诊断正确率为99.67%,运算时间为0.296 9s。新方法的诊断正确率比SVM和BPNN分别高出14.34%和9.34%,在线序列超限学习机的运算时间仅为SVM和BPNN的1/1 011和1/132。因此,该文所提方法适用于大数据量样本和多数据维度下的蒸汽冷却型燃料电池系统在线故障诊断。
服务
把本文推荐给朋友
加入我的书架
加入引用管理器
E-mail Alert
RSS
作者相关文章
刘嘉蔚
李奇
陈维荣
余嘉熹
燕雨
关键词 在线序列超限学习机蒸汽冷却型燃料电池系统故障诊断主成分分析数据驱动    
Abstract:In order to solve the fault diagnosis problem of evaporatively cooled fuel cell system, a novel fast fault diagnosis method for evaporatively cooled fuel cell system was proposed based on online sequential extreme learning machine and principal component analysis. The principal component analysis was used to filter the redundant information and obtain the fault eigenvectors that could reflect the states of evaporatively cooled fuel cell system. The fault feature vectors were classified by online sequential extreme learning machine, thereby effectively improving the diagnostic accuracy of the model and reducing the computation time. The case analysis shows that the novel method can quickly identify three healthy states of membrane drying failure, hydrogen leakage failure, and normal state. The diagnostic accuracy of the algorithm is 99.67% and the computation time is 0.296 9 seconds. The diagnostic accuracy of the novel method is 14.34% and 9.34% higher than those of SVM and BPNN respectively, and the computation time of online sequential extreme learning machine is only 1/1 011 of SVM and 1/132 of BPNN. Therefore, the proposed method is suitable for online fault diagnosis of evaporatively cooled fuel cell system with large data samples and multi-data dimensions.
Key wordsOnline sequential extreme learning machine    evaporatively cooled fuel cell system    fault diagnosis    principal component analysis    data-driven   
收稿日期: 2018-06-16      出版日期: 2019-09-26
PACS: TM911.48  
基金资助:国家自然科学基金(61473238)和四川省科技计划(应用基础面上项目)(19YYJC0698)资助项目
通讯作者: 李 奇 男,1984年生,教授,博士生导师,研究方向为轨道交通新能源技术、新能源并网发电技术、微电网运行与控制技术。E-mail: liqi0800@163.com   
作者简介: 刘嘉蔚 男,1993年生,博士研究生,研究方向为燃料电池健康诊断与控制技术。E-mail: daben@my.swjtu.edu.cn
引用本文:   
刘嘉蔚, 李奇, 陈维荣, 余嘉熹, 燕雨. 基于在线序列超限学习机和主成分分析的蒸汽冷却型燃料电池系统快速故障诊断方法[J]. 电工技术学报, 2019, 34(18): 3949-3960. Liu Jiawei, Li Qi, Chen Weirong, Yu Jiaxi, Yan Yu. Fast Fault Diagnosis Method of Evaporatively Cooled Fuel Cell System Based on Online Sequential Extreme Learning Machine and Principal Component Analysis. Transactions of China Electrotechnical Society, 2019, 34(18): 3949-3960.
链接本文:  
https://dgjsxb.ces-transaction.com/CN/10.19595/j.cnki.1000-6753.tces.181051          https://dgjsxb.ces-transaction.com/CN/Y2019/V34/I18/3949