基于改进LSTM的小电流系统HIF诊断方法  

HIF Diagnosis Based on Improved LSTM

作　　者：刘勇 吴伟丽 刘俊 陈保香 LIU Yong;WU Weili;LIU Jun;CHEN Baoxiang

机构地区：[1]安徽正广电电力技术有限公司,安徽合肥231200 [2]西安科技大学电气与控制工程学院,陕西西安710054

出　　处：《电力系统装备》2023年第7期35-37,共3页Electric Power System Equipment

摘　　要：文章针对小电流系统弧光高阻接地故障(High Impedance﹣Grounding Fault,HIF)信号微弱、监测样本数据偏少造成的诊断困难问题,提出一种基于改进LSTM的小电流系统弧光高阻接地故障辨识方法。针对电弧信号具有的零歇特征,采用卷积神经网络(CNN)对零歇局部特征加以提取,并利用LSTM进行全局特征提取,以最大限度地保留和凸显弧光接地故障信号特征,实现对小电流系统弧光高阻接地故障的准确辨识。并利用仿真数据和实测数据对方法加以验证,表明了方法的可行性和优越性。For the arc high resistance grounding fault of low current system(High Impedance﹣Grounding Fault,HIF)diagnostic difficulty caused by weak signal and little monitoring sample data,A fault identification method of small current system based on improved LSTM:Convolutional neural network(CNN),And perform the global feature extraction using LSTM,To maximize and highlight the arc grounding fault signal features,Realize the accurate identification of high resistance arc light grounding fault of low current system,Using simulation data and measured data,The feasibility and superiority of the method are shown.

关 键 词：CNN﹣LSTM HIF 故障诊断 

分 类 号：TM76[电气工程—电力系统及自动化]