基于脉冲电流和紫外脉冲的多源局部放电诊断方法  

Multi-source partial discharge detection method based on pulse current method and ultraviolet pulse method

作　　者：苏志雄 孙康 丁浩 张周胜 SU Zhixiong;SUN Kang;DING Hao;ZHANG Zhousheng(College of Electrical Engineering,Shanghai University of Electric Power,Shanghai 200090,China;State Grid Yancheng Dafeng District Power Supply Company,Yancheng 224100,China)

机构地区：[1]上海电力大学电气工程学院,上海200090 [2]国网盐城市大丰区供电公司,江苏盐城224100

出　　处：《绝缘材料》2024年第1期101-108,共8页Insulating Materials

基　　金：上海市科学技术委员会项目(20020500800、21DZ 2205000)。

摘　　要：本研究提出了一种基于脉冲电流法和紫外脉冲法联合检测的多源局部放电诊断方法,针对4种基本缺陷模型,搭建了开关柜多源局部放电实验平台。对联合检测得到的局部放电信息特征进行提取并构建数据库,使用k近邻(KNN)算法和有向无环图支持向量机(DAG-SVMs)算法对局部放电类型进行识别。结果表明:脉冲电流法测得的放电图谱中放电次数和放电量的变化与模型中缺陷种类有关,气隙、沿面缺陷的引入使放电量增大和图谱对称性提高,电晕缺陷的引入使放电次数增多;紫外脉冲法测得的放电脉冲数与多源放电模型中缺陷个数和缺陷的紫外/可见光比值有关,缺陷个数和紫外/可见光比值越高,放电脉冲数越大;KNN算法识别准确率最高可达到99.67%。In this paper,a multi-source partial discharge diagnosis method based on joint detection of pulse current method and ultraviolet pulse method was presented.A multi-source partial discharge experimental platform of switchgear was constructed for four basic defect models.The characteristics of partial discharge information obtained by joint detection were extracted and a database was constructed,and then the discharge types were identified by k-nearest neighbor(KNN)algorithm and directed acyclic graph SVMs(DAG-SVMs)algorithm.The results show that the change of discharge number and discharge quantity in discharge pattern measured by pulse current method are related to the defect type in model.The existence of void and surface defects will increase the discharge quantity and pattern symmetry,and the existence of corona defect will increase the discharge number.The discharge pulse number measured by ultraviolet pulse method is related to the defect number in multisource discharge model and the ratio of ultraviolet to visible light.The higher the number of defects and the ratio of ultraviolet to visible light,the larger the discharge pulse number.The recognition accuracy of the KNN algorithm can reach up to 99.67%.

关 键 词：开关柜 多源放电 KNN DAG-SVMs 脉冲电流法 紫外脉冲法 

分 类 号：TM855[电气工程—高电压与绝缘技术]