分布式光伏运维逆变器的直流注入故障检测  

DC Injection Fault Detection of Distributed Photovoltaic Operation&Maintenance Inverter

作　　者：黄力哲 王伟 杜闯 HUANG Li-zhe;WANG Wei;DU Chuang(Huaneng Renewables Corporation Limited,Beijing 100080,China)

机构地区：[1]华能新能源股份有限公司,北京100080

出　　处：《计算机仿真》2024年第12期127-131,共5页Computer Simulation

基　　金：华能新能源股份有限公司光伏管理系统计划项目(HN-62A1-202300135-JGQT00015)。

摘　　要：在分布式光伏运维中,当逆变器发生直流注入故障时,其产生的故障信号非常微弱,故障信号的稀疏性使得传统方法难以全面捕捉故障特征,从而影响故障检测的准确性和可靠性。为了有效确保光伏系统的安全稳定运行,提出一种分布式光伏运维逆变器直流注入故障实时检测方法。通过对分布式光伏运维中的逆变器内部故障电流特性展开分析,输出负序电流特征信息。在获取逆变器负序电流特征的基础上,将故障状态下稀疏测量点所关联的节点负序电压方程与压缩感知理论融合,即使在有限的数据点下也能准确重构故障特征,从而实现故障的快速定位。在此基础上,针对逆变器直流注入故障,提取三相均值电压作为故障检测依据,构建故障特征向量。将上述特征向量作为训练极限学习机(Extreme Learning Machine,ELM)故障检测模型的输入数据,通过离线训练过程形成故障分类器,将故障分类器输入到在线诊断流程中,最终实现故障检测。实验结果表明,所提方法可以获取精准的逆变器直流注入故障实时检测结果,有效保证逆变器的稳定工作。In distributed photovoltaic operation and maintenance,when the inverter has a DC injection fault,the fault signal generated is very weak.The sparsity of this fault signal makes it difficult for traditional methods to comprehensively capture the fault characteristics,thus affecting the accuracy and reliability of fault detection.In order to effectively ensure the safe and stable operation of photovoltaic systems,a real-time detection method for DC injection fault of distributed photovoltaic operation and maintenance inverter is proposed.By analyzing the internal fault current characteristics of the inverter in the distributed photovoltaic operation and maintenance,the characteristic information of negative sequence current is output.On the basis of acquiring the characteristics of inverter negative sequence current,the node negative sequence voltage equation associated with sparse measurement points in fault state is fused with the compression sensing theory,so that the fault characteristics can be accurately reconstructed even under limited data points,thus realizing rapid fault location.On this basis,the three-phase mean voltage is extracted as the fault detection basis for the inverter DC injection fault,and the fault feature vector is constructed.The above feature vector is used as the input data of the training extreme learning machine(ELM)fault detection model,and the fault classifier is formed through the offline training process.The fault classifier is input into the online diagnosis process,and finally the fault detection is realized.Experimental results show that the proposed method can obtain accurate real-time detection results of inverter DC injection faults,and effectively ensure the stable operation of the inverter.

关 键 词：分布式 光伏运维逆变器 直流注入故障 实时检测 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]