摘要
针对电力通信网故障类型复杂,故障诊断方式需要依靠人工经验,难以准确定位多链路故障的问题,基于深度学习技术,提出一种CNN卷积神经网络的电力通信故障诊断模型。首先,根据样本集大小,设计了CNN电力通信网故障诊断模型。然后,对采样的电力数据进行预处理,得到故障状态矩阵和故障标签输入模型训练,实现了故障诊断。最后,通过对模型参数学习率、最小批次大小、迭代次数调优,获得最优模型参数,并将最优模型与贝叶斯算法比较,证明所设计的模型准确率高达99.23%。
In view of the complex types of faults in the power communication network,the fault diagnosis methods need to rely on manual experience,and it is difficult to accurately locate the multi-link faults.Based on deep learning,this research proposes a fault diagnosis model of CNN convolutional neural network.Firstly,according to the size of the sample set,a CNN power communication network fault diagnosis model is designed.Secondly,the sampled power data are preprocessed to obtain the fault state matrix and the fault label input model training,and the fault diagnosis is realized.Finally,by tuning the model parameter learning rate,the minimum batch size,and the number of iterations,the optimal model parameters are obtained,and the optimal model is compared with the Bayesian algorithm,which proves that the accuracy of the model designed in this study is as high as 99.23%.
作者
邵淦
吴昊
姚朔晔
SHAO Gan;WU Hao;YAO Shuoye(State Grid Ningbo Electric Power Supply Company,Ningbo 315000,China)
出处
《微型电脑应用》
2022年第5期111-115,119,共6页
Microcomputer Applications
关键词
电力通信网
卷积神经网络
故障诊断
power communication network
convolution neural network
fault diagnosis
作者简介
邵淦(1988-),男,硕士,工程师,研究方向为电力通信;吴昊(1996-),男,本科,助理工程师,研究方向为电力通信;姚朔晔(1992-),男,本科,助理工程师,研究方向为电力通信。
