首页|期刊导航|电瓷避雷器|分布式光伏接入对配电网接地故障影响分析及其故障辨识研究

分布式光伏接入对配电网接地故障影响分析及其故障辨识研究OA

Influence of Distributed Photovoltaic Access on Grounding Faults in Distribution Network and Its Fault Identification

中文摘要英文摘要

大量的分布式光伏接入改变了传统配电网结构和潮流,使配电网接地故障特征发生改变.通过分析光伏并网容量、位置、光伏出力波动性及并网高次谐波对接地故障的零序电压的影响,发现对比光伏接入前,零序电压的各类时频域特征受到了不同程度的影响,甚至出现了部分接地故障时频域特征混淆的问题,造成故障辨识难度加大.对此,文中提出基于零序电压的VMD-MPE-CNN接地故障辨识算法.首先,采用VMD-MPE算法对零序电压信号进行分解,并计算本征模分量的多尺度排列熵得到故障特征矩阵.然后,构建CNN结构并利用各类故障样本特征矩阵进行训练,最终形成基于VMD-MPE-CNN接地故障分类辨识模型.实验表明,基于VMD-MPE算法的排列熵特征提取可有效差异化各类故障特征,对比传统时频特征辨识模型,文中模型的故障辨识准确率得到了提高.

A large number of distributed photovoltaic access changes the traditional distribution network structure and power flow,and changes the grounding fault characteristics of the distribution network.By analyzing the influence of grid-connected capacity,location,photovoltaic output fluctuation and grid-connected high harmonics on zero-sequence voltage of grounding faults,it is found that various time-frequency domain characteristics of zero-sequence voltage are affected to different degrees,and even some ground faults are confused in time-frequency domain,which makes fault identification more diffi-cult.The ground fault identification algorithm of VMD-MPE-CNN based on zero sequence voltage is proposed.Firstly,the zero-sequence voltage signal is decomposed by VMD-MPE algorithm,and the multi-scale arrangement entropy of eigenmode components is calculated to obtain the fault characteristic matrix.Then,the CNN structure is constructed and the feature matrix of various fault samples is used for training,and finally the ground fault classification and identification model based on VMD-MPE-CNN is formed.Experiments show that the algorithm based on VMD-MPE can effectively differentiate various fault features,and compared with the traditional time-frequency feature identification model,the fault i-dentification accuracy of the proposed model is improved.

龙国华;李琼;鄢文清;吴义辉;邹阳;刘斌

国网江西省电力有限公司电力科学研究院,南昌 330006南昌航空大学信息工程学院,南昌 330063国网江西省电力有限公司,南昌 330077南昌航空大学信息工程学院,南昌 330063国网江西省电力有限公司电力科学研究院,南昌 330006南昌航空大学信息工程学院,南昌 330063

分布式光伏并网特征提取变分模态分解多尺度排列熵接地故障辨识

distributed photovoltaic grid connectionfeature extractionvariational modal decompositionmultiscale alignment entropygrounding fault identification

《电瓷避雷器》 2026 (2)

46-55,66,11

国家自然科学基金资助项目(编号:52267008)国家电网江西省电力有限公司科技项目(编号:52182023000L).Project supported by National Natural Science Foundation of China(No.52267008)State Grid Jiangxi Electric Power Co.,Ltd.Sci-ence and Technology Project(No.52182023000L).

10.16188/j.isa.1003-8337.2026.02.007

评论