电力电缆局部放电检测模式识别方法研究  

Research on Pattern Recognition Method for Partial Discharge Detection of Power Cable

作　　者：王作君 才英博 赵欣 尹梦涵 Wang Zuojun;Cai Yingbo;Zhao Xin;Yin Menghan(State Grid Harbin Electric Power Supply Company Cable Operation and Maintenance Center,Harbin,China;Comprehensive Service Class 7 of State Grid Harbin Power Supply Company,Harbin,China)

机构地区：[1]国网哈尔滨供电公司电缆运检中心,黑龙江哈尔滨 [2]国网哈尔滨供电公司综服七班,黑龙江哈尔滨

出　　处：《科学技术创新》2025年第10期47-50,共4页Scientific and Technological Innovation

摘　　要：探究了神经网络模型在电力电缆局部放电检测与类型识别中的应用效果。使用经验模态分解方法提取局部放电信号的特征参数,并基于能量系数计算出局部放电的特征能量,根据特征能量的不同判断局部放电的具体类型。选择局部损伤、线芯毛刺、绝缘受潮、绝缘气泡4类局部放电各80组数据,使用基于神经网络的局部放电检测模型进行验证,模型学习速度为0.01次/s,最大训练次数10000次,期望误差0.01。结果表明,基于神经网络的局部放电检测模型能准确区分不同类型的局部放电,平均识别率为90.0%。该方法能较为准确地识别电力电缆的局部放电故障,并判断所述的具体类型,为电力系统的运行维护提供了参考。The application effect of neural network model in partial discharge detection and type identification of power cables is explored.The characteristic parameters of partial discharge signal are extracted by empirical mode decomposition method,and the characteristic energy of partial discharge is calculated based on the energy coefficient.The specific type of partial discharge is determined according to the different characteristic energy.Select 80 groups of data of local damage,core burr,insulation moisture and insulation bubble,and use the partial discharge detection model based on neural network to verify the model.The learning speed of the model is 0.01 times/s,the maximum training times is 10000 times,and the expected error is 0.01.The results show that the PD detection model based on neural network can accurately distinguish different types of PD,and the average recognition rate is 90.0%.The method can identify the partial discharge fault of power cable accurately,and judge the specific type,and provide a reference for the operation and maintenance of power system.

关 键 词：电力电缆 局部放电检测 经验模态分解 绝缘缺陷 

分 类 号：TM762[电气工程—电力系统及自动化]