基于 CONMI-LSGC-Transformer 的串联电弧故障检测

涂超; 李玉芬; 张永安; 刘睿捷; 刘玲

doi:10.19783/j.cnki.pspc.251077

欢迎来到能源电力期刊网！

English Version 期刊登录

您当前的位置：

首页 >

文章列表页 >

基于 CONMI-LSGC-Transformer 的串联电弧故障检测

应用研究 | 更新时间：2026-07-27

- 基于 CONMI-LSGC-Transformer 的串联电弧故障检测
- Series arc fault detection based on CONMI-LSGC-Transformer
- 电力系统保护与控制 2026年54卷第13期
- 作者机构：
  
  国网甘肃省电力公司张掖供电公司,甘肃,张掖,734000
- 作者简介：
- 基金信息：
- DOI：10.19783/j.cnki.pspc.251077
  中图分类号：
- 纸质出版：2026
- 稿件说明：
移动端阅览
涂超, 李玉芬, 张永安, 等. 基于 CONMI-LSGC-Transformer 的串联电弧故障检测[J]. 电力系统保护与控制, 2026,54(13).

TU Chao, LI Yufen, ZHANG Yongan, et al. Series arc fault detection based on CONMI-LSGC-Transformer[J]. Power System Protection and Control, 2026, 54(13).
涂超, 李玉芬, 张永安, 等. 基于 CONMI-LSGC-Transformer 的串联电弧故障检测[J]. 电力系统保护与控制, 2026,54(13). DOI： 10.19783/j.cnki.pspc.251077.

TU Chao, LI Yufen, ZHANG Yongan, et al. Series arc fault detection based on CONMI-LSGC-Transformer[J]. Power System Protection and Control, 2026, 54(13). DOI： 10.19783/j.cnki.pspc.251077.

摘要

针对电力通信机房中设备串联故障电弧监测难问题，基于CONMI特征选择算法和Transformer模型提出一种CONMI-LSGC-Transformer串联故障电弧检测方法。首先，基于不同工况下采集的电流数据，从时域、频域和时频域等领域提取相关特征以全面表征故障电弧影响，并使用CONMI进行特征筛选。然后，使用非因果时间卷积网络(non-causal temporal convolutional network

NCTCN)分支和门控分支构造局部尺度门控卷积(local-scale gated convolution

LSGC)结构，并用该结构替换Transformer编码器中的前馈网络，构建LSGC-Transformer模型，实现全局特征依赖和局部细节相结合。最后，将筛选出的特征子集通过该模型进行故障电弧识别。经对比实验和消融实验验证，所提方法性能优良，识别准确率可达98.89%，证明了该方法在故障电弧检测中的稳定性和有效性。

Abstract

Aiming at the challenge of monitoring series arc faults in power communication equipment rooms

a CONMI-LSGC-Transformer series arc fault detection method is proposed based on the CONMI feature selection algorithm and Transformer model. First

based on the collected current data under different working conditions

relevant features are extracted from the time domain

frequency domain

and time-frequency domain

providing a comprehensive characterization of the effects of arc faults. The CONMI algorithm is then used for feature selection. Subsequently

the non-causal temporal convolutional network (NCTCN) branch and the gated branch are used to construct the local-scale gated convolution (LSGC) structure

which is used to replace the feedforward network in the Transformer encoder. The LSGC-Transformer model is constructed to realize the combination of global feature dependency learning and local detail extraction. Finally

the selected feature subset is fed into the proposed model for series arc fault identification. Comparative and ablation experiments demonstrate the superior performance of the proposed method

achieving a recognition accuracy of 98.89%

thereby confirming its stability and effectiveness in fault arc detection.

关键词

Keywords

references

GONG H, LI Y, ZHANG J, et al. A new filter feature selection algorithm for classification task by ensembling pearson correlation coefficient and mutual information[J]. Engineering Applications of Artificial Intelligence, 2024, 131.

TORRES M E, COLOMINAS M A, SCHLOTTHAUER G, et al. A complete ensemble empirical mode decomposition with adaptive noise[C]//IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), May 22-27, 2011, Prague, Czech Republic: 4144-4147.

LI T, ZHAO Q, JIA S, et al. Enhanced soil organic matter estimation using spectral feature selection with empirical mode decomposition and pearson correlation coefficients[J]. Infrared Physics Technology, 2025, 106.

ZHONG L, HUANG R, GAO L, et al. A novel variable selection method based on binning-normalized mutual information for multivariate calibration[J]. Molecules, 2023, 28(15): 56-72.

XIE Z, CHEN L, LI Y. Optimization of dam safety monitoring models based on residual compensation: a multidimensional temporal framework integrating TCN-BiLSTM-attention mechanism[J]. Engineering Structures, 2026, 348.

WANG C, FU Z, ZHANG Z, et al. Fault diagnosis of power transformer in one-key sequential control system of intelligent substation based on a transformer neural network model[J]. Processes, 2024, 12(4).

WU G, XU J, ZHANG Y, et al. Weighted feature fusion of dual attention convolutional neural network and transformer encoder module for ocean HABs classification[J]. Expert Systems with Applications, 2024, 243.

XU W, DAI W, LI D, et al. Short-term wind power prediction based on a variational mode decomposition-BiTCN-Transformer hybrid model[J]. Energies, 2024, 17(16).

浏览量

下载量

CSCD

文章被引用时，请邮件提醒。

提交

工具集

关联资源

基于改进 FRWT 的柔性直流配电网故障保护方法

基于 TimesNet-Transformer 融合模型的电能质量时序预测方法

计及预案式失配冲击的响应驱动频率稳定紧急切负荷策略

柔性直流电网双端量自适应限流控制策略研究

基于EWT改进限流电抗电压的柔性直流配电网故障保护方法