配电线路易发生接地故障，当接地介质为沙地等非金属介质时，接地介质被烧灼产生电弧现象，因此被称为弧光高阻接地故障，其暂稳态电气量变化微弱、难以被常规线路保护装置识别，从而引发火灾、触电等重大安全事故。针对这一问题，配合日益成熟的人工智能对配电网弧光高阻接地故障辨识及选线进行了研究。
  首先，在分析配电网弧光高阻接地故障机理基础上，研究暂稳态电气量变化特征及其影响因素，对比中性点不同接地方式下弧光高阻接地故障零序电流和零序电压突变的特征，利用PSCAD搭建了1OkV配电网系统模型，模拟各种工况下配电网弧光高阻接地故障，获取了不同故障情况下零序电流及故障相电压波形数据，为后续故障辨识与选线提供数据基础。
  其次，针对弧光高阻接地故障信号微弱易受噪声干扰，存在实测样本数据偏少，特征提取因难的问题，提出基于加权融合格拉姆角场(Gramian Angular Field,.GAF)与改进卷积神经网络(Convolutional Neural Net-.workd,.CNN)的弧光高阻接地故障特征提取与识别方法。首先，对故障零序电流进行格拉姆角场编码，并将得到的格拉姆角和场与角差场加权平均融合成一张全信息空间域图像；其次，在传统卷积神经网络的全连接层处嵌入自适应权重的通道注意力模块，并将传统CNN模型中的softmax分类器用由粒子群优化的支持向量机替换；最后，在不同条件下验证所提辨识方法的可行性和鲁棒性。

  最后，针对弧光高阻接地故障存在暂态信号特征辨识度低，选线判据容易受其他故障条件干扰，选线准确率低的问题，提出基于优化变分模态分解(Variational Mode Decomposition,,VMD)与改进CNN的故障选线方法。首先，将健全线路与故障线路零序电流作为VMD输入，并用经过融合鱼鹰算法与柯西变异优化的麻雀算法(Osprey-Cauchy-Sparrow Search Algorithm,OCSSA)对VMD参数寻优并得到最佳IMF分解量，计算最佳MF的参数指标并以此作为选线特征量：其次，在CNN全连接层前嵌入BiLSTM,形成弧光高阻接地故障选线模型；最后，在不同条件下验证所提选线方法的可行性和鲁棒性。

Distribution lines are prone to grounding faults. When the grounding medium is non-metallic medium such as sand, the grounding medium is burned to produce arc phenomenon. Therefore, it is called arc high-resistance grounding fault. Its transient and steady-state electrical quantity changes weakly and is difficult to be identified by conventional line protection devices, which leads to major safety accidents such as fire and electric shock.To address this issue, research has been conducted on the identification and line selection of high-resistance arc grounding faults in distribution networks, leveraging the increasingly mature technology of artificial intelligence.

Firstly, based on the analysis of the mechanism of high-resistance arc grounding faults in distribution networks, the characteristics of transient and steady-state electrical quantity changes and their influencing factors were studied. By comparing the features of zero-sequence current and zero-sequence voltage mutations during high-resistance arc grounding faults under different grounding methods of neutral points, a 10kV distribution network system model was built using PSCAD to simulate high-resistance arc grounding faults in distribution networks under various working conditions. Zero-sequence current and fault phase voltage waveform data under different fault conditions were obtained, providing a data basis for subsequent fault identification and line selection.

Furthermore, addressing the issues of weak and noisy signals, limited practical sample data, and difficulty in feature extraction associated with high-resistance arc grounding faults, a method for feature extraction and identification based on the weighted fusion of the Gramian Angular Field (GAF) and an improved Convolutional Neural Network (CNN) is proposed. Firstly, the fault zero-sequence current is encoded using the GAF, and the resulting Gramian Angle Sum Field and Angle Difference Field are weighted and averaged to create a comprehensive information spatial domain image. Secondly, an adaptive weight channel attention module is embedded in the fully connected layer of the traditional CNN, and the softmax classifier in the traditional CNN model is replaced with a Particle Swarm Optimization (PSO)-optimized Support Vector Machine (SVM). Finally, the feasibility and robustness of the proposed identification method are verified under different conditions.

Finally, to address the issues of low transient signal feature identification, susceptibility to interference from other fault conditions, and low accuracy in line selection for high resistance arc grounding faults, a fault line selection method based on optimized variational mode decomposition (VMD) and improved CNN is proposed. Firstly, the zero sequence currents of the healthy and faulty lines are used as inputs for VMD, and the Osprey Cauchy Sparrow Search Algorithm (OCSSA), which combines the Osprey algorithm and Cauchy mutation optimization, is used to optimize the VMD parameters and obtain the optimal IMF decomposition amount. The optimal IMF parameter index is calculated and used as the line selection feature; Secondly, BiLSTM is embedded in front of the CNN fully connected layer to form a high resistance arc grounding fault line selection model; Finally, verify the feasibility and robustness of the proposed line method under different conditions.

[1] 叶远波, 黄太贵, 谢民, 等. 小电阻接地系统高阻和间歇性弧光接地故障继电保护研究[J]. 电网与清洁能源, 2021, 37(09): 9-17+26.

[2] Amin G, Ali H M, L. H G, et al. High-Impedance Fault Detection in the Distribution Network Using the Time-Frequency-Based Algorithm[J]. IEEE Transactions on Power Delivery, 2015, 30(03): 1260-1268.

[3] 王宾, 倪江, 王海港, 等. 输电线路弧光高阻接地故障单端测距分析[J]. 中国电机工程学报, 2017, 37(05): 1333-1341.

[4] Chen J, Phung T, Blackburn T , et al. Detection of high impedance faults using current transformers for sensing and identification based on features extracted using wavelet transform[J]. IET Generation, Transmission Distribution, 2016, 10(12): 2990-2998.

[5] 韦明杰, 石访, 张恒旭, 等. 基于零序电流波形区间斜率曲线的配电网高阻接地故障检测[J]. 电力系统自动化, 2020, 44(14): 164-171.

[6] 陈筱薷. 基于暂态零序电流投影的谐振接地系统高阻接地故障检测技术[D]. 青岛: 中国石油大学(华东), 2017.

[7] Baqui I, Zamora I, Mazón J, et al. High impedance fault detection methodology using wavelet transform and artificial neural networks[J]. Electric Power Systems Research, 2011, 81(07): 1325-1333.

[8] Liu Y J, Chen C I, Fu W C, et al. A hybrid approach for low-voltage AC series arc fault detection[J]. Energies, 2023, 16(03): 1256.

[9] 和敬涵, 罗国敏, 程梦晓, 等. 新一代人工智能在电力系统故障分析及定位中的研究综述[J]. 中国电机工程学报, 2020, 40(17): 5506-5516.

[10] Wei M, Shi F, Zhang H, et al. Wideband synchronous measurement-based detection and location of high impedance fault for resonant distribution systems with integration of DERs[J]. IEEE Transactions on Smart Grid, 2022, 14(02): 1117-1134.

[11] Souza F P, Silveira L F Q, Costa F B, et al. High-impedance fault identification using cyclostationary characteristic analysis[J]. Electric Power Systems Research, 2021, 195: 107-150.

[12] Brito N S D, de Souza B A, dos Santos W C, et al. Analysis of the influence of the window used in the Short-Time Fourier Transform for High Impedance Fault detection[C]//2016 17th International Conference on Harmonics and Quality of Power (ICHQP). IEEE, 2016: 350-355.

[13] Wontroba A, Morais A P, Cardoso G J, et al. High-impedance fault detection on downed conductor in overhead distribution networks[J]. Electric Power Systems Research, 2022, 211: 108-216.

[14] Soheili A, Sadeh J, Bakhshi R. Modified FFT based high impedance fault detection technique considering distribution non-linear loads: Simulation and experimental data analysis[J]. International Journal of Electrical Power & Energy Systems, 2018, 94: 124-140.

[15] Wang Y, Zhang J, Yao C, et al. A mathematical method for local defects and faults identification of 10 kV three‐core cable based on input impedance spectrum[J]. IET Science, Measurement & Technology, 2022, 16(08): 467-478.

[16] Sun J, Zhang X, Wei S, et al. High-impedance fault location method of three-phase distribution lines based on electromagnetic time reversal[J]. Electric Power Systems Research, 2022, 212: 108-629.

[17] 韦明杰, 石访, 张恒旭, 等. 基于零序电流波形区间斜率曲线的配电网高阻接地故障检测[J]. 电力系统自动化, 2020, 44(14): 164-171.

[18] 郭霖徽, 刘亚东, 王鹏, 等. 基于相空间重构与平均电导特征的配电网单相接地故障辨识[J]. 电力系统自动化, 2019, 43(07): 192-198.

[19] 刘志文, 董旭柱, 邹林, 等. 基于零序电流衰减周期分量的高阻接地故障区段定位[J].电力系统自动化, 2020, 44(07): 161-168.

[20] 杨帆, 金鑫, 沈煜, 等. 小电流接地故障多层次分类及其识别模型[J]. 电力系统自动化, 2018, 42(15): 186-191.

[21] 杨帆, 刘鑫星, 沈煜, 等. 基于零序电流投影系数的小电阻接地系统高阻接地故障保护[J]. 电网技术, 2020, 44(03): 1128-1133.

[22] 薛永端, 李娟, 陈筱薷, 等. 谐振接地系统高阻接地故障暂态选线与过渡电阻辨识[J].中国电机工程学报, 2017, 37(17): 5037-5048+5223.

[23] 高锋阳, 李昭君, 袁成, 等. 含特殊负荷的配电网故障定位与识别[J]. 电力自动化设备, 2020, 40(08): 96-108.

[24] Lala H, Karmakar S. Detection and experimental validation of high impedance arc fault in distribution system using empirical mode decomposition[J]. IEEE Systems Journal, 2020, 14(03): 3494-3505.

[25] Biswal M, Ghore S, Malik O P, et al. Development of time-frequency based approach to detect high impedance fault in an inverter interfaced distribution system[J]. IEEE Transactions on Power Delivery, 2021, 36(06): 3825-3833.

[26] 刘科研, 叶学顺, 李昭, 等. 基于多分辨率小波变换的配电网高阻接地故障检测方法[J].高电压技术, 2023, 49(10): 4247-4256.

[27] Wang X, Gao J, Wei X, et al. Faulty Feeder Detection Under High Impedance Faults for Resonant Grounding Distribution Systems[J]. IEEE Transactions on Smart Grid, 2022, 14(03): 1880-1895.

[28] Bhatnagar M, Yadav A, Swetapadma A. Fuzzy based relaying scheme for high impedance faults in DG integrated distribution system[J]. Electric Power Systems Research, 2022, 202: 107602

[29] Lopes G N, Menezes T S, Santos G G, et al. High impedance fault detection based on harmonic energy variation via S-transform[J]. International Journal of Electrical Power & Energy Systems, 2022, 136: 107681.

[30] 崔朴奕, 李国丽, 张倩, 等. 基于VMD-CNN的小电流接地系统故障电弧检测方法研究[J]. 电力系统保护与控制, 2021, 49(23): 18-25.

[31] Gomes D P S, Ozansoy C, Ulhaq A, et al. The effectiveness of different sampling rates in vegetation high-impedance fault classification[J]. Electric Power Systems Research, 2019, 174: 105872.

[32] Gomes D P S, Ozansoy C, Ulhaq A. High-sensitivity vegetation high-impedance fault detection based on signal's high-frequency contents[J]. IEEE Transactions on Power Delivery, 2018, 33(03): 1398-1407.

[33] Shiyuan W, Payman D. On the Use of Artificial Intelligence for High Impedance Fault Detection and Electrical Safety[J]. IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, 2020, 56(06): 7208-7216

[34] Krishna B C, Anamika Y, Mohammad P. An Intelligent Detection of High-Impedance Faults for Distribution Lines Integrated With Distributed Generators[J]. IEEE Systems Journal, 2020, 14(01): 870-879.

[35] 刘科研, 詹惠瑜, 谭云耀, 等. 基于改进极限学习机的配电网高阻接地故障检测方案[J].电网技术, 2023, 47(05): 1886-1899.

[36] 薛永端, 李娟, 陈筱薷, 等. 谐振接地系统高阻接地故障暂态选线与过渡电阻辨识[J].中国电机工程学报, 2017, 37(17): 5037-5048+5223.

[37] 王宾, 崔鑫, 董新洲. 配电线路弧光高阻故障检测技术综述[J]. 中国电机工程学报, 2020, 40(01): 96-107+377.

[38] 曾晓丹. 基于深度学习的配电网单相接地故障选线方法研究[D]. 福建: 福州大学, 2018.

[39] 潘超, 陈祥, 蔡国伟, 等. 基于小波包尺度-能量占比的变压器三相不平衡绕组振动特征辨识[J]. 仪器仪表学报, 2020, 41(04): 129-137.

[40] Ye X, Hu Y, Shen J, et al. An adaptive optimized TVF-EMD based on a sparsity-impact measure index for bearing incipient fault diagnosis[J]. IEEE Transactions on Instrumentation and Measurement, 2020, 70: 1-11.

[41] Guo Y, Zhang Z. Generalized variational mode decomposition: A multiscale and fixed-frequency decomposition algorithm[J]. IEEE Transactions on Instrumentation and Measurement, 2021, 70: 1-13.

[42] Rehman U N, Aftab H. Multivariate Variational Mode Decomposition[J]. IEEE Transactions on Signal Processing, 2019, 67(23): 6039-6052.

[43] 王志坚, 常雪, 王俊元, 等. 排列熵优化改进变模态分解算法诊断齿轮箱故障[J]. 农业工程学报, 2018, 34(23): 59-66.

[44] Sedighi A-R, Haghifam M-R, Malik O P, et al. High im⁃pedance fault detection based on wavelet transform and statistica l pattern recognition[J]. IEEE Trans on Power De⁃livery, 2005, 20(04): 2414-2421.

[45] Gonzalez C, Tant J, Germain J G, et al. Directional, high-impedance fault detection in isolated neutral distribution grids[J]. IEEE Transactions on Power Delivery, 2018, 33(05): 2474-2483.

[46] 刘渝根, 王建南, 马晋佩, 等. 结合小波包变换和5次谐波法的谐振接地系统综合故障选线方法[J]. 高电压技术, 2015, 41(05): 1519-1525.

[47] 张淑清, 翟欣沛, 董璇, 等. EMD及Duffing振子在小电流系统故障选线方法中的应用[J]. 中国电机工程学报, 2013, 33(10): 161-167+5.

[48] 刘柱揆, 曹敏, 董涛. 基于波形相似度的小电流接地故障选线[J]. 电力系统保护与控制, 2017, 45(21): 89-95.

[49] 张淑清, 马跃, 李盼, 等. 基于改进的广义谐波小波包分解和混沌振子的小电流接地系统故障选线[J]. 电工技术学报, 2015, 30(03): 13-20+43.

[50] Xiaowei W, Jie G, Xiangxiang W, et al. Single Line to Ground Fault Detection in a Non-Effectively Grounded Distribution Network[J]. IEEE Transactions on Power Delivery, 2018, 33(06): 3173-3186.

[51] Xiaowei W, Jie G, Mingfei C, et al. Faulty Line Detection Method Based on Optimized Bistable System for Distribution Network[J]. IEEE Transactions on Industrial Informatics, 2018, 14(04): 1370-1381.

[52] 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.

[53] Yuanyuan W, Jiaming Z, Zewen L, et al. Discriminant-Analysis-Based Single-Phase Earth Fault Protection Using Improved PCA in Distribution Systems[J]. IEEE Transactions on Power Delivery, 2015, 30(04): 1974-1982.

[54] 李纳纳. 基于机器学习的配电网弱故障选线方法[D]. 淄博: 山东理工大学, 2021.

[55] Krizhevsky A, Sutskever I, Hinton G. Image Net Classification with Deep Convolutional Neural Networks[C]// NIPS. Curran Associates Inc. 2012: 1097-1105.

[56] He K, Zhang X, Ren S, et al. Deep Residual Learning for Image Recognition[C]//IEEE Conference on Computer Vision&Pattern Recognition. IEEE Computer Society. 2016: 770-778.

[57] Aytar Y, Vondrick C, Torralba A. Sound Net: Learning Sound Representations from Unlabeled Video[C]//Annual Conference on Neural Information Processing Systems. 2016: 892-900.

[58] Guo Sheng, Yang Tao, Gao Wei. A Novel Fault Diagnosis Method for Rotating Machinery Based on a Convolutional Neural Network[J]. Sensors, 2018: 18(05):1429.

[59] Esteva A, Kuprel B, Novoa R A, et al. Dermatologist-level classification of skincancer with deep neural networks[J]. Nature, 2017, 542(7639): 115-118.

[60] 刘志文, 董旭柱, 邹林, 等. 基于零序电流衰减周期分量的高阻接地故障区段定位[J].电力系统自动化, 2020, 44(07): 161-168.

[61] 徐丙垠, 薛永端, 冯光, 等. 配电网接地故障保护若干问题的探讨[J]. 电力系统自动化, 2019, 43(20): 1-7.

[62] 张淑清, 杜灵韵, 王册浩, 等. 基于格拉姆角场与改进CNN-ResNet的风电功率预测方法[J]. 电网技术, 2023, 47(04): 1540-1548.

[63] 郭威. 配电线路故障类型辨识及故障选线定位方法的研究[D]. 北京: 华北电力大学, 2022.

[64] 刘俊奇, 涂文轩, 祝恩. 图卷积神经网络综述[J]. 计算机工程与科学, 2023, 45(08): 1472-1481.

[65] 江泽涛, 秦嘉奇, 张少钦. 参数池化卷积神经网络图像分类方法[J]. 电子学报, 2020, 48(09): 1729-1734.

[66] 冉鹏, 王灵, 李昕, 等. 改进Softmax分类器的深度卷积神经网络及其在人脸识别中的应用[J]. 上海大学学报(自然科学版), 2018, 24(03): 352-366.

[67] 占可, 王寅杰, 董路南, 等. 基于改进格拉姆角场和注意力的滚动轴承故障诊断[J/OL]. 轴承, 2023, 8(20): 1-8.

[68] 尹文哲, 夏虹, 彭彬森, 等. 基于CNN-SVM的核电厂轴承故障诊断方法[J]. 哈尔滨工程大学学报, 2023, 44(03): 410-417.

[69] Rodriguez J D, Perez A, Lozano J A. Sensitivity analysis of k-fold cross validation in prediction error estimation. IEEE transactions on pattern analysis and machine intelligence, 2010, 32(03): 569-575.

[70] Kumar A, Zhou Y, Xiang J. Optimization of VMD using kernel-based mutual information for the extraction of weak features to detect bearing defects[J]. Measurement, 2021, 168: 108402.

[71] 刘恒勇, 史帅彬, 徐旭辉, 等. 一种关联RNN模型的非侵入式负荷辨识方法[J]. 电力系统保护与控制, 2019, 47(13): 162-170.

[72] Chen K, Huo Q. Training deep bidirectional LSTM acoustic model for LVCSR by a context-sensitive-chunk BPTT approach[J]. IEEE/ACM Transactions on Audio, Speech and Language Processing (TASLP), 2016, 24(07): 1185-1193.