计算机与数字工程2025，Vol.53Issue(1)：284-289,6.DOI:10.3969/j.issn.1672-9722.2025.01.050
双通道混合神经网络电弧故障检测方法
Arc Fault Detection Method for Double Channel Hybrid Neural Network
摘要
Abstract
Aiming at the problem that the current series arc detection algorithm relies on human analysis and insufficient fea-ture extraction,a two-channel hybrid neural network arc fault detection method is proposed.Firstly,an arc data acquisition plat-form is built to analyze the typical load waveform.Then a model of two parallel network channels is built to achieve efficient feature extraction,one of which is a convolutional neural network that introduces residual modules,Leaky ReLU functions,and CBAM modules to extract local features in the data space,and the other channel is to extract global time series features through the GRU network.Experimental tests show that the proposed model can accurately identify arc faults,which is better than other typical mod-els.关键词
双通道/电弧故障/残差模块/时序特征Key words
dual channel/arc fault/residual module/timing characteristics分类
信息技术与安全科学引用本文复制引用
黄冬梅,张玲,孙锦中,胡安铎..双通道混合神经网络电弧故障检测方法[J].计算机与数字工程,2025,53(1):284-289,6.基金项目
上海市科委地方院校能力建设项目(编号:20020500700)资助. （编号:20020500700）
