基于极限学习机的光伏逆变器软故障辨识方法  

Soft Fault Identification Method of Photovoltaic Inverter Based on Extreme Learning Machine

作　　者：李蓉[1] 肖家平[1] LI Rong;XIAO Jiaping(Huainan Vocational and Technical College,Huainan Anhui 232001,China)

机构地区：[1]淮南职业技术学院,安徽淮南232001

出　　处：《重庆科技学院学报（自然科学版）》2020年第3期83-87,108,共6页Journal of Chongqing University of Science and Technology：Natural Sciences Edition

基　　金：2018年度安徽省高校自然科学研究重点项目“煤矿井下大容量锂电池状态估计与健康管理关键技术研究”(KJ2018A0759)。

摘　　要：针对光伏逆变器软故障的诊断问题,提出了一种基于极限学习机的逆变器电容退化程度诊断方法。提取逆变器电容信号的统计特征参数作为故障特征,利用极限学习机实现电容参数的辨识,确定逆变器的电容软故障。实验结果表明,基于故障信号的统计特征,运用极限学习机实现特征参数辨识的方法,具有抗干扰性强、诊断速度快、诊断精度高等优势,而且检测信号易获取,诊断成本低,适合用于在线诊断。Aiming at the problem of diagnosing soft faults of photovoltaic inverters,a diagnosis method of inverter capacity degradation based on extreme learning machine is proposed.The statistical characteristic parameters of the inverter capacitor signal are extracted as the fault characteristics,and the capacitance parameters of the inverter are identified by the extreme learning machine to determine the capacitor's soft fault.The experimental results show that,based on the statistical characteristics of the fault signal,the method of using extreme learning machines to realize the feature parameter identification has the advantages of strong anti-interference,fast diagnosis speed,high diagnosis accuracy,etc.And the detection signal is easy to obtain and the diagnosis cost is low,so it is suitable for on-line diagnosis.

关 键 词：光伏逆变器 故障诊断 电容 统计特征 极限学习机 

分 类 号：TM464[电气工程—电器]