基于DNN-GRU的导轨式电能表故障诊断方法  

DNN-GRU based fault diagnosis method for rail-mounted energy meters

作　　者：梁海涛 陈腾飞 马燕红 赵雨濛 LIANG Haitao;CHEN Tengfei;MA Yanhong;ZHAO Yumeng(Wuzhong Power Supply Company,State Grid Ningxia Electric Power Corporation,Ningxia Wuzhong 751100,China)

机构地区：[1]国网宁夏电力有限公司吴忠供电公司,宁夏吴忠751100

出　　处：《自动化与仪器仪表》2025年第3期329-334,共6页Automation & Instrumentation

基　　金：宁夏电力有限公司科技项目资助(5229WZ23 0007)。

摘　　要：针对导轨式电能表发生故障将严重影响电网公司效益及用户日常生活的问题,准确对导轨式电能表故障类型进行诊断、及时修复故障电能表,对保证电力系统正常运行有着重要的意义。为了有效利用导轨式电能表在运行过程中产生的大量高维历史数据,提出了一种融合深度神经网络和门控循环单元(DNN-GRU)的神经网络方法,用于导轨式电能表故障数据的多维分析和故障类型诊断。首先,使用DNN增强网络的特征提取能力,解决历史数据中多种特征数据和高维问题。之后,将处理后的特征信息输入到GRU门控循环单元中,利用GRU对非线性数据的强大处理能力,解决历史数据集中数据弱相关问题,并提高故障诊断的准确率。算例研究结果表明,DNN-GRU方法能够有效减少训练迭代次数,缩短训练时间,并提高诊断准确性。In view of the problem that the failure of the guide rail type electric energy meter will seriously affect the benefit of the power grid company and the daily life of the users,it is of great significance to accurately diagnose the fault type of the guide rail type electric energy meter and repair the fault electric energy meter in time to ensure the normal operation of the power system.In order to effectively utilize the large amount of high-dimensional historical data generated during the operation of the rail-type electric energy meter,this paper proposes a neural network method that combines deep neural network and gated recurrent unit( DNN-GRU) for multi-dimensional analysis and fault type diagnosis of rail-type electric energy meter fault data.Firstly,DNN is used to enhance the feature extraction ability of the network and solve the problem of multiple feature data and high dimension in historical data.After that,the processed feature information is input into the GRU gated recurrent unit,and the strong processing ability of GRU to nonlinear data is used to solve the problem of weak correlation of data in historical data set and improve the accuracy of fault diagnosis.The results of the example study show that the DNN-GRU method can effectively reduce the number of training iterations,shorten the training time,and improve the diagnostic accuracy.

关 键 词：深度神经网络 门控循环单元 故障诊断 导轨式电能表 

分 类 号：TM933.4[电气工程—电力电子与电力传动]