姜凯华,张永,谢迎谱,李特,王少华,雷梦飞,王博闻
JIANG Kaihua,ZHANG Yong,XIE Yingpu,LI Te,WANG Shaohua,LEI Mengfei,WANG Bowen

• 1:国网浙江省电力有限公司电力科学研究院
• 2:国网浙江省电力有限公司
• 3:国网电科院武汉南瑞有限公司
• 4:电网雷击风险预防湖北省重点实验室

摘要(Abstract)：

输电线路故障原因复杂多样，且各类故障的发生频率存在显著差异，导致故障暂态波形数据在类别上存在明显不平衡性。为提高故障识别的准确性并增强模型的泛化能力，提出一种结合深度森林与SMOTE（合成少数过采样技术）算法的输电线路故障识别方法。首先，提取故障暂态波形的时频域特征，并将其输入深度森林模型中以获取各类故障的预测概率；随后，针对输出的类别概率分布，采用SMOTE算法进行过采样，缓解少数类样本在预测阶段的判别偏差，从而提升整体识别精度。通过与多种传统机器学习算法的对比实验验证，所提方法在故障识别准确率和泛化性能方面均表现出更优性能。
Power transmission line faults exhibit multi-causal origins and imbalanced occurrence frequencies across fault types, resulting in significant class imbalance in transient waveform data. To improve fault identification accuracy and enhance model generalization capability, this paper proposes an integrated approach combining Deep Forest and synthetic minority oversampling technique(SMOTE). The method first extracts time-frequency domain features from fault transient waveforms and inputs them into the Deep Forest model to obtain prediction probabilities for each fault type. The SMOTE is then applied to the output probability distribution to perform oversampling, mitigating discrimination bias against minority-class samples during prediction and thereby improving overall identification accuracy. Comparative experiments with various traditional machine learning algorithms demonstrate that the proposed method achieves better performance in both fault identification accuracy and generalization capability.

关键词(KeyWords)： 深度森林;过采样;故障识别;输电线路;机器学习
Deep Forest;oversampling;fault identification;transmission line;machine learning

Abstract:

Keywords:

基金项目(Foundation): 国网浙江省电力有限公司科技项目（5211DS23000M）

作者(Author): 姜凯华,张永,谢迎谱,李特,王少华,雷梦飞,王博闻
JIANG Kaihua,ZHANG Yong,XIE Yingpu,LI Te,WANG Shaohua,LEI Mengfei,WANG Bowen

DOI: 10.19585/j.zjdl.202512012

参考文献(References)：

• [1]徐铭铭，冯光，张林林，等.基于同步波形的配电网故障诊断技术综述[J].电力工程技术，2019,38(5):138-146.XU Mingming,FENG Guang,ZHANG Linlin,et al.Distribution network fault diagnosis technology based on synchronous waveforms[J].Electric Power Engineering Technology,2019,38(5):138-146.
• [2]霍红刚，周蠡，蔡杰，等.基于先验知识Faster R-CNN的输电线路无人机图像识别方法[J].智慧电力，2024,52(6):108-115.HUO Honggang,ZHOU Li,CAI Jie,et al. UAV image recognition method for transmission line based on prior knowledge faster R-CNN[J].Smart Power,2024,52(6):108-115.
• [3]董桓毓.基于故障录波数据特征的输电线路故障原因辨识[D].北京：华北电力大学，2022.DONG Huanyu. Identification of transmission line fault causes based on the characteristics of fault recording data[D]. Beijing:North China Electric Power University,2022.
• [4]张永东，陶伟叶，常欣丽，等.基于数学形态学与希尔伯特-黄变换的线路故障测距方法研究[J].山东电力技术，2024,51(12):78-83.ZHANG Yongdong,TAO Weiye,CHANG Xinli,et al.Research on line fault ranging method based on the mathematical morphology and Hilbert-Huang transform[J].Shandong Electric Power,2024,51(12):78-83.
• [5]刘沐辰，安景革，程定一，等.基于双端同步响应的高压输电线路故障定位方法[J].智慧电力，2023,51(12):15-22.LIU Muchen,AN Jingge,CHENG Dingyi,et al.Fault location method for HV transmission line based on twoterminal synchronous response[J]. Smart Power,2023,51(12):15-22.
• [6]孙晓茜.柔性直流输电线路故障识别与测距技术研究[D].郑州：郑州大学，2021.SUN Xiaoqian. Research on fault identification and location technology of MMC-HVDC transmission lines[D].Zhengzhou:Zhengzhou University,2021.
• [7]苏超，杨强.基于多视图稀疏特征选择的架空输电线路故障原因判别[J].智慧电力，2023,51(3):96-103.SU Chao,YANG Qiang.Fault cause identification of overhead transmission line based on multi-view sparse feature selection[J].Smart Power,2023,51(3):96-103.
• [8]黄明伟，王力，周仕豪，等.基于暂态能量比的交流输电线路故障选相方法[J].智慧电力，2023,51(2):98-104.HUANG Mingwei,WANG Li,ZHOU Shihao,et al.Fault phase selection method for AC transmission lines based on transient energy ratio[J]. Smart Power,2023,51(2):98-104.
• [9]汤同峰，王峰，蔡德胜.基于小波分析的输电线路故障识别检测[J].电子设计工程，2024,32(1):73-76.TANG Tongfeng,WANG Feng,CAI Desheng. Fault identification and detection of transmission lines based on wavelet analysis[J].Electronic Design Engineering,2024,32(1):73-76.
• [10]缪希仁，林瑞聪.基于关联维数与极端学习机的高压输电线路雷击过电压故障识别[J].高电压技术，2016,42(5):1519-1526.MIAO Xiren,LIN Ruicong. Lightning over-voltage fault identification of high-voltage transmission line based on correlation dimension and extreme learning machine[J].High Voltage Engineering,2016,42(5):1519-1526.
• [11]姜泽苗，袁喆.基于机器学习算法的电力系统故障诊断[J].电气技术与经济，2023(9):366-368.
• [12]赵岩，孙江山.基于贝叶斯优化随机森林输电线路故障识别方法[J].中国新技术新产品，2024(8):10-12.
• [13]张尔康，杨岸.基于VMD和PSO-SVM的输电线路故障诊断[J].兰州文理学院学报（自然科学版），2023,37(6):67-71.ZHANG Erkang,YANG An.Transmission line fault diagnosis based on VMD and PSO-SVM[J].Journal of Lanzhou University of Arts and Science(Natural Sciences),2023,37(6):67-71.
• [14]裴东锋，刘勇，闫柯柯，等.一种基于CNN与FFT-ELM的输电线路故障识别与定位方法[J].电力科学与技术学报，2024,39(1):164-170.PEI Dongfeng,LIU Yong,YAN Keke,et al. A method based on CNN and FFT-ELM for fault identification and location of transmission lines[J].Journal of Electric Power Science and Technology,2024,39(1):164-170.
• [15]沈银，席燕辉，陈子璇.基于多通道卷积双向长短时记忆网络的输电线故障分类[J].电力系统保护与控制，2022,50(3):114-120.SHEN Yin,XI Yanhui,CHEN Zixuan.Transmission line fault classification based on MCCNN-BiLSTM[J].Power System Protection and Control,2022,50(3):114-120.
• [16]WANG W,YU B,SUN H,et al.Weak feature fault identification and location of distribution network based on multi-task learning[C]//2022 5th International Conference on Renewable Energy and Power Engineering(REPE). September 28-30,2022,Beijing,China:IEEE,2022:173-178.
• [17]石洪波，陈雨文，陈鑫.SMOTE过采样及其改进算法研究综述[J].智能系统学报，2019,14(6):1073-1083.SHI Hongbo,CHEN Yuwen,CHEN Xin.Summary of research on SMOTE oversampling and its improved algorithms[J]. CAAI Transactions on Intelligent Systems,2019,14(6):1073-1083.
• [18]赵锦阳，卢会国，蒋娟萍，等.一种非平衡数据分类的过采样随机森林算法[J].计算机应用与软件，2019,36(4):255-261.ZHAO Jinyang,LU Huiguo,JIANG Juanping,et al. An oversampling random forest algorithm for classification of imbalance data[J].Computer Applications and Software,2019,36(4):255-261.
• [19]周玉，孙红玉，房倩，等.不平衡数据集分类方法研究综述[J].计算机应用研究，2022,39(6):1615-1621.ZHOU Yu,SUN Hongyu,FANG Qian,et al.Review of imbalanced data classification methods[J].Application Research of Computers,2022,39(6):1615-1621.
• [20]姚长元，罗国敏.基于置信网络的直流输电线路暂态信号识别方法[J].浙江电力，2019,38(6):3-8.YAO Changyuan,LUO Guomin.Transient signal identification method for DC transmission line based on belief network[J].Zhejiang Electric Power,2019,38(6):3-8.
• [21]石万宇，魏军强，赵云灏.基于改进麻雀算法-支持向量机的输电线路故障诊断[J].浙江电力，2021,40(11):38-45.SHI Wanyu,WEI Junqiang,ZHAO Yunhao. Transmission line fault diagnosis based on support vector machine optimized by improved sparrow search algorithm[J].Zhejiang Electric Power,2021,40(11):38-45.
• [22]袁丹，王谊，李伟明，等.基于分类模型的配电线路故障研判方法研究[J].浙江电力，2018,37(2):11-15.YUAN Dan,WANG Yi,LI Weiming,et al.Research on fault diagnosis method for distribution line based on classification model[J].Zhejiang Electric Power,2018,37(2):11-15.
• [23]孙文成，李健，彭宇辉，等.基于样本不均衡和特征优选多源融合的输电线路故障类型辨识[J].电测与仪表，2024,61(12):79-89.SUN Wencheng,LI Jian,PENG Yuhui,et al. Transmission line fault type identification based on the sample imbalance and feature preferred multi-source fusion[J].Electrical Measurement&Instrumentation,2024,61(12):79-89.
• [24]ZHOU Z H,FENG J.Deep forest[J].National Science Review,2019,6(1):74-86.
• [25]国家质量监督检验检疫总局，中国国家标准化管理委员会.输电线路分布式故障诊断系统：GB/T 35721—2017[S].北京：中国标准出版社，2018.
• [26]韩晓慧，杜松怀，苏娟，等.触电信号暂态特征提取及故障类型识别方法[J].电网技术，2016,40(11):3591-3596.HAN Xiaohui,DU Songhuai,SU Juan,et al. Fault transient feature extraction and fault type identification for electrical shock signals[J]. Power System Technology,2016,40(11):3591-3596.