燕山大学学报2025，Vol.49Issue(2)：137-145,9.DOI:10.3969/j.issn.1007-791X.2025.02.006
基于自适应ECA的串联故障电弧检测
Series arc fault detection based on adaptive ECA
摘要
Abstract
In the case of increasing and complex load on the user side,the series fault arc becomes more difficult to identify effectively,which threatens the safety of the line and the operation of the system.Previous detection methods mostly use two-dimensional or three-dimensional models,which usually require more computation than one-dimensional models,and pay less attention to the information between feature channels.Therefore,a series fault arc detection method based on adaptive Efficient Channel Attention(ECA)is proposed to adaptively extract the effective features in the current signal.Firstly,the one-dimensional convolutional network is improved by using efficient channel attention and residual structure,and the network model structure is constructed.Then,through the built arc fault experimental platform,various load current data under normal working conditions and arc fault conditions are collected,and the corresponding database is established.Finally,the constructed one-dimensional network model is used to train and classify the data samples,so that it can effectively identify the arc fault.The results show that the average recognition accuracy of the model for fault arc is 98.68%,which has a good recognition effect.关键词
串联故障电弧/高效通道注意力/残差结构/卷积神经网络/电弧检测Key words
series arc fault/efficient channel attention/residual structure/convolutional neural network/arc detection分类
动力与电气工程引用本文复制引用
袁建华,黄淘,卢云..基于自适应ECA的串联故障电弧检测[J].燕山大学学报,2025,49(2):137-145,9.基金项目
湖北省自然科学基金资助项目(2020CFB248) （2020CFB248）
