基于注意力机制-卷积神经网络的配电网单相接地故障选线方法

doi:10.12204/j.issn.1000-7229.2023.04.010

电力建设 ›› 2023, Vol. 44 ›› Issue (4): 82-93.doi: 10.12204/j.issn.1000-7229.2023.04.010

基于注意力机制-卷积神经网络的配电网单相接地故障选线方法

陈池瑶¹^,²(), 苗世洪¹^,²(), 殷浩然¹^,²(), 王子欣¹^,²(), 韩佶¹^,²()

1.强电磁工程与新技术国家重点实验室(华中科技大学),武汉市 430074
2.电力安全与高效湖北省重点实验室(华中科技大学),武汉市 430074

收稿日期:2022-07-07 出版日期:2023-04-01 发布日期:2023-03-30
通讯作者: 苗世洪 E-mail:m202171885@hust.edu.cn;shmiao@hust.edu.cn;hr_yin@126.com;zixinwang98@hust.edu.cn;han_ji1993@163.com
作者简介:陈池瑶(1999),女,硕士研究生,主要研究方向为人工智能在电力系统中的应用,E-mail:m202171885@hust.edu.cn;
殷浩然(1997),男,硕士研究生,主要研究方向为人工智能在电力系统中的应用,E-mail:hr_yin@126.com;
王子欣(1998),男,硕士研究生,主要研究方向为电力系统设备故障识别及电力系统规划,E-mail:zixinwang98@hust.edu.cn;
韩佶(1993),男,博士研究生,主要研究方向为新能源技术与人工智能技术,E-mail:han_ji1993@163.com。
基金资助:
国家电网有限公司总部科技项目(SGHADK00PJJS2000026)

Single-Phase Grounding-Fault Line Selection Method Based on Attention Mechanism-Convolution Neural Network for Distribution Network

CHEN Chiyao¹^,²(), MIAO Shihong¹^,²(), YIN Haoran¹^,²(), WANG Zixin¹^,²(), HAN Ji¹^,²()

1. State Key Laboratory of Advanced Electromagnetic Engineering and Technology (Huazhong University of Science and Technology), Wuhan 430074, China
2. Hubei Electric Power Security and High Efficiency Key Laboratory (Huazhong University of Science and Technology), Wuhan 430074, China

Received:2022-07-07 Online:2023-04-01 Published:2023-03-30
Contact: MIAO Shihong E-mail:m202171885@hust.edu.cn;shmiao@hust.edu.cn;hr_yin@126.com;zixinwang98@hust.edu.cn;han_ji1993@163.com
Supported by:
Science and Technology Project of State Grid Corporation of China(SGHADK00PJJS2000026)

摘要/Abstract

摘要：

针对配电网发生单相接地故障的电气量特征难以提取,故障选线准确性与鲁棒性难以保证,给电网安全稳定运行带来巨大隐患问题,提出一种基于注意力(attention)机制-卷积神经网络(convolutional neural networks, CNN)的配电网单相接地故障选线方法。首先,采用S变换将故障各线路时序零序电流数据转化为CNN可识别的二维输入矩阵;其次,在传统CNN分类算法进行故障选线的基础上引入注意力机制,建立了基于Attention-CNN的故障选线新模型;最后,通过仿真数据与工程实际数据验证模型的选线结果,并将所提Attention-CNN模型与同类CNN优化方法在不同干扰条件下的性能进行对比。结果表明,所提Attention-CNN模型能够大幅提高选线效率,减少迭代次数,实现更高效、准确的故障选线,具有较强的实际应用价值。

关键词: 单相接地故障, 配电网, 卷积神经网络(CNN), 注意力机制, 模糊集理论, S变换, 深度学习

Abstract:

In case of distribution network single-phase grounding fault, characteristics are difficult to capture, which may affect the selection correctness and thus bringing hidden dangers to the safe and stable operation of the power grid. Hence, a new method based on attention mechanism and convolutional neural network (CNN) is proposed. Firstly, zero-sequence current data is preprocessed by S-transform. Secondly, Attention-CNN model is established by introducing attention mechanism to CNN. Finally, performance of the proposed model is verified by both simulation and real grid data, and compared with other methods under different fault conditions. Results show that the proposed Attention-CNN model can accomplish more efficient and accurate selection, as well as wide application.

Key words: single-phase grounding fault, distribution network, convolutional neural network (CNN), attention mechanism, fuzzification theory, S transform, deep learning

中图分类号:

TM71

陈池瑶, 苗世洪, 殷浩然, 王子欣, 韩佶. 基于注意力机制-卷积神经网络的配电网单相接地故障选线方法[J]. 电力建设, 2023, 44(4): 82-93.

CHEN Chiyao, MIAO Shihong, YIN Haoran, WANG Zixin, HAN Ji. Single-Phase Grounding-Fault Line Selection Method Based on Attention Mechanism-Convolution Neural Network for Distribution Network[J]. ELECTRIC POWER CONSTRUCTION, 2023, 44(4): 82-93.

图/表 19

图1 配电网单相接地故障电流分布

Fig.1 Distribution of single-phase grounding fault current in distribution network

图2 小电流接地系统单相接地故障零序等值电路

Fig.2 Equivalent circuit of single-phase grounding fault in small-current grounding system

图3 Attention-CNN模型结构

Fig.3 Structure of Attention-CNN

图4 Attention-CNN模型池化层原理

Fig.4 Structure of Attention-CNN pooling layer

表1 不同结构下Attention-CNN模型训练结果

Table 1 Training results of Attention-CNN model with different structures

表2 Attention-CNN模型训练参数

Table 2 Parameters of Attention-CNN model

图5 输入样本特征矩阵构建过程

Fig.5 Construction of characteristic sampling matrix

图6 3种数据预处理方法选线性能对比

Fig.6 Comparison of three preprocessing methods

图7 不同消弧线圈补偿度下故障零序电流波形

Fig.7 Zero-sequence current waveform under different compensation degree of arc suppression coil

表3 测试集混淆矩阵元素

Table 3 Elements in testing set confusion matrix

图8 Attention-CNN与模糊集-CNN模型训练结果对比

Fig.8 Results of Attention-CNN and Fuzzification-CNN

表4 零序电流采样不同步时各方法选线正确率对比

Table 4 Comparison of CNN methods under s ampling asynchroneity

图9 模型选线抗噪性能结果对比

Fig.9 Anti-noise ability comparison of CNN models

图10 模型选线误差与迭代次数关系曲线

Fig.10 Relationship between model error and iteration

表5 20 dB噪声下模型选线准确率与迭代次数关系对比

Table 5 Comparison of methods with different iterations under 20 dB noise environment

图A1 仿真模型拓扑

Fig.A1 Topology of simulation model

图A2 某实际电网拓扑

Fig.A2 Topology of real power grid

表A1 仿真模型线路参数

Table A1 Line parameters of simulation model

表A2 仿真模型故障遍历表

Table A2 Fault traversal table of simulation model

参考文献 28

[1]	邵文权, 程畅, 卫晓辉, 等. 利用暂态电流Hausdorff距离的谐振配电网故障选线方案[J]. 电力系统保护与控制, 2022, 50(8): 33-42.
	SHAO Wenquan, CHENG Chang, WEI Xiaohui, et al. Fault line selection scheme using the Hausdorff distance of transient current in resonant distribution networks[J]. Power System Protection and Control, 2022, 50(8): 33-42.
[2]	严欢, 胡俊杰, 黄旦莉, 等. 考虑电动汽车虚拟电厂灵活性和高比例光伏接入的配电网规划[J]. 电力建设, 2022, 43(11): 14-23. doi: 10.12204/j.issn.1000-7229.2022.11.002
	YAN Huan, HU Junjie, HUANG Danli, et al. Distribution network planning considering the flexibility of EV virtual power plants and high penetration of PV[J]. Electric Power Construction, 2022, 43(11): 14-23. doi: 10.12204/j.issn.1000-7229.2022.11.002
[3]	魏科文, 张靖, 何宇, 等. 基于VMD和相关性聚类的谐振接地系统单相接地故障选线[J]. 电力系统保护与控制, 2021, 49(22): 105-113.
	WEI Kewen, ZHANG Jing, HE Yu, et al. Single-phase grounding fault line selection in a resonant grounding system based on VMD and correlation clustering[J]. Power System Protection and Control, 2021, 49(22): 105-113.
[4]	王卫, 宫成, 曹文远, 等. 交直流配电网交流侧非对称故障情况下直流侧电压波动特性分析与控制[J]. 电力建设, 2021, 42(11): 90-99. doi: 10.12204/j.issn.1000-7229.2021.11.010
	WANG Wei, GONG Cheng, CAO Wenyuan, et al. Analysis and control of DC side voltage fluctuation characteristics under unbalanced fault of AC side in AC/DC distribution network[J]. Electric Power Construction, 2021, 42(11): 90-99. doi: 10.12204/j.issn.1000-7229.2021.11.010
[5]	郭凤仪, 高洪鑫, 唐爱霞, 等. 局部二值模式直方图匹配的串联故障电弧检测及选线[J]. 电工技术学报, 2020, 35(8): 1653-1661.
	GUO Fengyi, GAO Hongxin, TANG Aixia, et al. Series arc fault detection and line selection based on local binary pattern histogram matching[J]. Transactions of China Electrotechnical Society, 2020, 35(8): 1653-1661.
[6]	郭清滔, 吴田. 小电流接地系统故障选线方法综述[J]. 电力系统保护与控制, 2010, 38(2): 146-152.
	GUO Qingtao, WU Tian. Survey of the methods to select fault line in neutral point ineffectively grounded power system[J]. Power System Protection and Control, 2010, 38(2): 146-152.
[7]	LEE I W C, DASH P K. S-transform-based intelligent system for classification of power quality disturbance signals[J]. IEEE Transactions on Industrial Electronics, 2003, 50(4): 800-805. doi: 10.1109/TIE.2003.814991 URL
[8]	林湘宁, 高艳, 刘沛, 等. 基于零序补偿导纳的小电流接地系统单相故障保护新方法[J]. 中国电机工程学报, 2006, 26(10): 45-49.
	LIN Xiangning, GAO Yan, LIU Pei, et al. A novel method to identify the single phase-to-earth fault in the neutral un-effectual grounded system using the zero-sequence compensated admittance[J]. Proceedings of the CSEE, 2006, 26(10): 45-49.
[9]	李卫华. 小波神经网络在小电流接地系统单相接地故障定位中的应用研究[D]. 郑州: 华北水利水电大学, 2013.
	LI Weihua. Application of wavelet neural network in single-phase grounding fault location of small current grounding system[D]. Zhengzhou: North China University of Water Resources and Electric Power, 2013.
[10]	杨月, 孙博, 马晓忱, 等. 融合注意力机制的输电线路故障概率预测模型[J]. 电力建设, 2022, 43(3): 42-49. doi: 10.12204/j.issn.1000-7229.2022.03.005
	YANG Yue, SUN Bo, MA Xiaochen, et al. Prediction model of transmission line fault probability applying attention mechanism[J]. Electric Power Construction, 2022, 43(3): 42-49. doi: 10.12204/j.issn.1000-7229.2022.03.005
[11]	王尊贤, 武守远, 栾晓明. 基于虚拟能量变化率的谐振接地系统高阻接地故障选线方法[J]. 电力系统自动化, 2022, 46(3): 137-146.
	WANG Zunxian, WU Shouyuan, LUAN Xiaoming. Fault line selection method for high impedance grounding fault of resonant grounded system based on rate of change of virtual energy[J]. Automation of Electric Power Systems, 2022, 46(3): 137-146.
[12]	JING G, RU L. A new wavelet packet method of single-phase earth fault line selection in distribution network based on the maximum difference comparison[C]// 2009 International Conference on Electrical Machines and Systems. IEEE, 2010: 1-5.
[13]	张伟刚, 张保会, 胡海松, 等. 应用小波包分析实现配电网单相接地故障选线[J]. 电力系统自动化, 2009, 33(23): 60-64.
	ZHANG Weigang, ZHANG Baohui, HU Haisong, et al. Application of wavelet packet analysis in phase-to-ground fault detection of distribution networks[J]. Automation of Electric Power Systems, 2009, 33(23): 60-64.
[14]	徐玉琴, 张婧婷, 边华跃. 复杂配电网系统单相接地故障选线算法[J]. 电测与仪表, 2019, 56(1): 22-27.
	XU Yuqin, ZHANG Jingting, BIAN Huayue. An algorithm for earth fault line selection in complex distribution network system[J]. Electrical Measurement & Instrumentation, 2019, 56(1): 22-27.
[15]	张国栋, 蒲海涛, 刘凯. 基于深度学习的小电流接地系统故障选线方法[J]. 发电技术, 2019, 40(6): 548-554. doi: 10.12096/j.2096-4528.pgt.19103
	ZHANG Guodong, PU Haitao, LIU Kai. Fault line selection method of small current grounding system based on deep learning[J]. Power Generation Technology, 2019, 40(6): 548-554. doi: 10.12096/j.2096-4528.pgt.19103
[16]	LIANG J F, JING T J, NIU H N, et al. Two-terminal fault location method of distribution network based on adaptive convolution neural network[J]. IEEE Access, 2020, 8: 54035-54043. doi: 10.1109/Access.6287639 URL
[17]	殷浩然, 苗世洪, 郭舒毓, 等. 基于S变换相关度和深度学习的配电网单相接地故障选线新方法[J]. 电力自动化设备, 2021, 41(7): 88-96.
	YIN Haoran, MIAO Shihong, GUO Shuyu, et al. Novel method for single-phase grounding fault line selection in distribution network based on S-transform correlation and deep learning[J]. Electric Power Automation Equipment, 2021, 41(7): 88-96.
[18]	白星振, 宋昭杉, 葛磊蛟, 等. 含分布式电源的复杂配电网相间故障定位等效解耦模型[J]. 电力建设, 2022, 43(2): 45-53. doi: 10.12204/j.issn.1000-7229.2022.02.006
	BAI Xingzhen, SONG Zhaoshan, GE Leijiao, et al. An equivalent decoupling model for fault location in complex distribution network with distributed generation[J]. Electric Power Construction, 2022, 43(2): 45-53. doi: 10.12204/j.issn.1000-7229.2022.02.006
[19]	罗毅, 田吉华. 基于双端数据和算法融合的输电线路故障测距新方法[J]. 电力系统及其自动化学报, 2016, 28(12): 1-6.
	LUO Yi, TIAN Jihua. Transmission line fault location method based on double-ended data and algorithm-fusion[J]. Proceedings of the CSU-EPSA, 2016, 28(12): 1-6.
[20]	王子馨, 胡俊杰, 刘宝柱. 基于长短期记忆网络的电力系统量测缺失数据恢复方法[J]. 电力建设, 2021, 42(5): 1-8. doi: 10.12204/j.issn.1000-7229.2021.05.001
	WANG Zixin, HU Junjie, LIU Baozhu. Recovery method for missing measurement data of power systems based on long short-term memory networks[J]. Electric Power Construction, 2021, 42(5): 1-8. doi: 10.12204/j.issn.1000-7229.2021.05.001
[21]	刘守亮, 肖先勇, 杨洪耕. 基于S变换模时频矩阵相似度的短时电能质量扰动分类[J]. 电网技术, 2006, 30(5): 67-71.
	LIU Shouliang, XIAO Xianyong, YANG Honggeng. Classification of short duration power quality disturbance based on module time-frequency matrixes similarity by S-transform[J]. Power System Technology, 2006, 30(5): 67-71.
[22]	KONG Z, TANG B, DENG L, et al. Condition monitoring of wind turbines based on spatio-temporal fusion of SCADA data by convolutional neural networks and gated recurrent units[J]. Renewable Energy, 2020, 146: 760-768. doi: 10.1016/j.renene.2019.07.033 URL
[23]	WANG F, JIANG M Q, QIAN C, et al. Residual attention network for image classification[C]// 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). IEEE, 2017: 6450-6458.
[24]	阴登辉. 基于扩容消弧线圈的10 kV系统单相接地故障补偿方法研究[D]. 沈阳: 沈阳化工大学, 2021.
	YIN Denghui. Research on compensation method of single-phase grounding fault in 10 kV system based on expanded arc suppression coil[D]. Shenyang: Shenyang University of Chemical Technology, 2021.
[25]	喻锟, 胥鹏博, 曾祥君, 等. 基于模糊测度融合诊断的配电网接地故障选线[J]. 电工技术学报, 2022, 37(3): 623-633.
	YU Kun, XU Pengbo, ZENG Xiangjun, et al. Grounding fault line selection of distribution networks based on fuzzy measures integrated diagnosis[J]. Transactions of China Electrotechnical Society, 2022, 37(3): 623-633.
[26]	庄胜斌, 缪希仁, 江灏, 等. 基于改进欧氏-动态时间弯曲距离的谐振接地配电网单相高阻接地故障选线方法[J]. 电网技术, 2020, 44(1): 273-281.
	ZHUANG Shengbin, MIAO Xiren, JIANG Hao, et al. A line selection method for single-phase high-impedance grounding fault in resonant grounding system of distribution network based on improved euclidean-dynamic time warping distance[J]. Power System Technology, 2020, 44(1): 273-281.
[27]	谭悦. 基于Attention机制的CNN-BiLSTM股价预测模型及量化交易策略研究[D]. 哈尔滨: 哈尔滨工业大学, 2021.
	TAN Yue. Research on CNN-BiLSTM stock price forecast model and quantitative trading strategy based on attention mechanism[D]. Harbin: Harbin Institute of Technology, 2021.
[28]	QIAN L D, HU L B, ZHAO L, et al. Sequence-dropout block for reducing overfitting problem in image classification[J]. IEEE Access, 2020, 8: 62830-62840. doi: 10.1109/Access.6287639 URL

基于注意力机制-卷积神经网络的配电网单相接地故障选线方法

Single-Phase Grounding-Fault Line Selection Method Based on Attention Mechanism-Convolution Neural Network for Distribution Network

RichHTML

PDF

可视化

摘要/Abstract

引用本文

使用本文

图/表 19

参考文献 28

相关文章 15

编辑推荐

Metrics

本文评价

[1]	崔文倩, 魏军强, 赵云灏, 高炜, 陈康. 双碳目标下含重力储能的配电网多目标运行优化[J]. 电力建设, 2023, 44(4): 45-53.
[2]	范馨予, 黄媛, 吴疆, 孙增杰, 宁静波. 考虑源网荷储协同优化的配电网韧性提升策略[J]. 电力建设, 2023, 44(4): 63-73.
[3]	纪鹏志, 李光肖, 王琳, 刘思贤, 刘宗杰, 马梓耀, 王照琪, 唐巍. 基于Transformer深度学习网络的主动配电网多元源荷灾损辨识方法[J]. 电力建设, 2023, 44(3): 56-65.
[4]	谭媛, 张文海, 王杨. 基于多源信息融合的配网故障外部原因识别[J]. 电力建设, 2023, 44(3): 77-84.
[5]	于昊正, 赵寒杰, 李科, 朱星旭, 皇甫霄文, 樊江川, 李翠萍, 李军徽. 计及需求响应的分布式光伏集群承载能力评估[J]. 电力建设, 2023, 44(2): 122-130.
[6]	王婷, 陈晨, 谢海鹏. 配电网对分布式电源和电动汽车的承载力评估及提升方法综述[J]. 电力建设, 2022, 43(9): 12-24.
[7]	杨秀, 焦楷丹, 孙改平, 陈小毅, 杜佳玮, 仇志鑫. 考虑负荷多无功用电场景的城市配电网无功优化配置[J]. 电力建设, 2022, 43(8): 42-52.
[8]	李振坤, 王艺龙, 季亮, 张智泉. 基于虚拟变电站的配电网薄弱区域集群双层规划研究[J]. 电力建设, 2022, 43(8): 53-65.
[9]	刘家妤, 沈冰, 周健, 杜洋, 冯煜尧, 唐敏, 陈韵含, 李佳旭, 王颖. 计及关键负荷功能恢复需求的韧性城市配电网恢复方法[J]. 电力建设, 2022, 43(8): 66-75.
[10]	李军徽, 马得轩, 朱星旭, 李翠萍, 侯涛. 基于ADMM算法的主动配电网分层优化经济调度[J]. 电力建设, 2022, 43(8): 76-86.
[11]	徐艳春, 张进, 汪平, MI Lu. 考虑配电网静态电压稳定性的微电网优化配置[J]. 电力建设, 2022, 43(8): 87-101.
[12]	宫建锋, 周宗川, 靳盘龙, 张斌, 韩一鸣, 冯雪. 考虑灵活性的交直流混合配电网储能双层规划[J]. 电力建设, 2022, 43(7): 48-56.
[13]	张桂红, 王琛, 张祥成, 王世斌, 白左霞, 张中锋. 一种适用于直流配电网中半桥-全桥混合型MMC的全电平逼近调制策略[J]. 电力建设, 2022, 43(6): 75-83.
[14]	徐加银, 汪涛, 王加庆, 刘琛, 马英浩. 考虑微电网灵活性资源属性的配电网规划方法[J]. 电力建设, 2022, 43(6): 84-92.
[15]	陈泽西, 王朴, 肖万芳, 贾东强, 耿欣扬, 杨立. 计及可靠性收益的配电网柔性多状态开关及分布式电源综合优化配置[J]. 电力建设, 2022, 43(6): 93-100.