基于多级特征提取的并网逆变器故障诊断策略  

Fault Diagnosis Strategy Based on Multi-level Feature Extraction for Grid-connected Inverter

作　　者：耿俊超 王天真[1] 韩金刚[1] 陈国栋 汤天浩[1] GENG Junchao;WANG Tianzhen;HAN Jingang;CHEN Guodong;TANG Tianhao(Logistics Engineering College,Shanghai Maritime University,Shanghai 201306,China;Technology Center of Shanghai Electric Power Transmission&Distribution Group,Shanghai 200042,China)

机构地区：[1]上海海事大学物流工程学院,上海201306 [2]上海电气输配电集团技术中心,上海200042

出　　处：《电源学报》2022年第4期28-36,共9页Journal of Power Supply

基　　金：国家自然科学基金资助项目(61203089)。

摘　　要：由于降低了谐波含量,减少了开关损耗,且易于模块化,级联型H桥多电平逆变器在可再生能源并网系统中得到了广泛的应用。随着电平数的增加,其使用的开关器件(IGBT)数量随之增多,相应的故障概率与故障类型也在增加。在多种故障情况下,传统的故障特征提取方法不能有效地提取故障的关键特征,从而造成诊断精度的下降。针对这一问题,提出了一种基于多级特征提取的故障诊断策略。首先利用PCA对故障特征进行初步提取,然后基于欧氏距离阈值提取关键故障特征,最后使用ELM分类器完成故障诊断。实验结果表明,所提出的诊断策略具有更少的输入特征向量维度和更高的故障诊断精度。Owing to its reduced harmonic content,reduced switching loss,and easy modularization,the cascaded Hbridge multilevel inverter has been widely applied to renewable energy grid-connected systems.However,as the number of levels increases,the number of switch devices(e.g.,IGBTs)grows,and the corresponding fault probability and fault types also increase.Under multiple fault conditions,the traditional fault feature extraction method cannot effectively extract the key features of fault,resulting in a decrease in the diagnosis accuracy.To solve this problem,a fault diagnosis strategy based on multi-level feature extraction is proposed.First,PCA is used to preliminarily extract fault features.Then,the key fault features are extracted based on the Euclidean distance threshold.Finally,the ELM classifier is used to complete the fault diagnosis.Experimental results show that the proposed diagnosis strategy has fewer input feature vector dimensions and higher fault diagnosis accuracy.

关 键 词：级联H桥多电平逆变器 并网系统 特征提取 故障诊断 

分 类 号：TM464[电气工程—电器]