基于FCM和SRF组合的光伏组件故障诊断研究  

Fault diagnosis of photovoltaic modules based on combination of FCM and SRF

作　　者：汪洋[1] 闫天一 陈凤云 肖文 胡春花[2] WANG Yang;YAN Tian-yi;CHEN Feng-yun;XIAO Wen;HU Chun-hua(School of Electrical and Information Engineering,Jiangsu University,Zhenjiang Jiangsu 212013,China;Department of Electrical and Information Science,Zhenjiang College,Zhenjiang Jiangsu 212028,China)

机构地区：[1]江苏大学电气信息工程学院,江苏镇江212013 [2]镇江高等专科学校电子与信息学院,江苏镇江212028

出　　处：《电源技术》2019年第12期2009-2013,2057,共6页Chinese Journal of Power Sources

基　　金：江苏省重点研发计划(社会发展)面上项目(BE2016727);镇江市重点研发(产业前瞻与共性关键技术)项目(GY2015018,GY2017026);镇江市重点研发计划(社会发展)项目(SH2016015)

摘　　要：为了在较长的时间段内准确诊断光伏组件的故障,提出了一种基于FCM (fuzzy C-means)算法和SRF算法组合的光伏组件故障诊断方法。利用FCM将样本数据按照最大隶属度原则进行相似日聚类并得到聚类中心,用每一类样本数据分别训练SRF分类模型,将上述聚类中心和SRF分类模型组成FCM-SRF分类模型来判断光伏组件的运行状态。通过实验证明了该方法适用于样本数据的时间和季节跨度较大的情况,其判断结果具有较高的准确性和有效性。In order to diagnose the faults of photovoltaic modules accurately in a large period of time,a combined fault diagnosis method based on fuzzy C-means clustering algorithm and SRF algorithm is proposed.Firstly,the sample data are clustered by FCM according to the principle of maximum membership degree and the clustering centers are obtained.Then the classification model of each sample data is established by SRF.Finally,the FCM-SRF classification model combined with the clustering centers and classification models is used to judge the operation status of photovoltaic modules.Experiments show that this method is suitable for the sample data with large time and seasonal span,and its judgment results have high accuracy and validity.

关 键 词：光伏组件 故障诊断 模糊C均值聚类 随机森林 STACKING 

分 类 号：TM914[电气工程—电力电子与电力传动]