基于改进神经网络的电压监测仪故障自动诊断方法  

Automatic fault diagnosis method of voltage monitor based on improved neural network

作　　者：刘娈 李超 LIU Luan;LI Chao(Dalian Power Supply Company,State Grid Liaoning Electric Power Co.,Ltd.,Dalian 116000,China)

机构地区：[1]国网辽宁省电力有限公司大连供电公司,辽宁大连116000

出　　处：《电子设计工程》2023年第10期167-171,共5页Electronic Design Engineering

摘　　要：为使电压监测仪母线端的纯电感负载保持较低数值水平,从而实现对电网故障行为的准确诊断与控制,提出基于改进神经网络的电压监测仪故障自动诊断方法。通过集成与优化处理神经网络的方式,判定故障样本空间的构成有效性,再联合自适应阈值结果,实现基于改进神经网络的电压故障类型定义。在此基础上,设置故障定位规则,根据已知的故障特征分析原理,实现对诊断参数的按需选取,完成基于改进神经网络电压监测仪故障自动诊断方法的设计与应用。对比实验结果表明,与改进堆叠型诊断手段相比,在改进神经网络作用下,电压监测仪母线端的纯电感负载作用得到有效控制,可满足准确诊断并控制电网故障行为的实际应用需求。In order to keep the pure inductive load at the busbar of the voltage monitor at a low value level,so as to realize the accurate diagnosis and control of the grid fault behavior,an automatic fault diagnosis method of the voltage monitor based on an improved neural network is proposed.By integrating and optimizing the processing of neural networks,the validity of the structure of the fault sample space is determined,and then the adaptive threshold results are combined to realize the definition of voltage fault types based on the improved neural network.On this basis,the fault location rules are set,based on the known fault characteristic analysis principle,the on-demand selection of diagnostic parameters is realized,and the design and application of the automatic fault diagnosis method based on the improved neural network voltage monitor is completed.The comparative experiment results show that,compared with the improved stacked diagnosis method,the pure inductive load at the busbar end of the voltage monitor can be effectively controlled under the effect of the improved neural network,which can meet the actual application requirements of accurately diagnosing and controlling the fault behavior of the power grid.

关 键 词：神经网络 电压监测仪 故障诊断 样本空间 自适应阈值 纯电感负载 

分 类 号：TN92[电子电信—通信与信息系统]