Partial Discharge Pattern Recognization Method for Generator Based on SURF-BoVF Features and Active Learning Model EQB-LightGBM

BI Guangjun; LONG Shixing; LUO Mingxing; WANG Hongyu; PENG Xiaosheng; QI Shiwei

doi:10.13296/j.1001-1609.hva.2025.10.017

您当前的位置：

首页 >

文章列表页 >

Partial Discharge Pattern Recognization Method for Generator Based on SURF-BoVF Features and Active Learning Model EQB-LightGBM

Digital Twin Technology and Applications for Multi-scenario in New Power System | 更新时间：2025-10-21

- Partial Discharge Pattern Recognization Method for Generator Based on SURF-BoVF Features and Active Learning Model EQB-LightGBM
- High Voltage Apparatus Vol. 61, Issue 10, Pages: 159-167(2025)
- 作者机构：
- 作者简介：
- 基金信息：
- DOI：10.13296/j.1001-1609.hva.2025.10.017
  CLC：
- Received：18 March 2025，
  
  Revised：2025-06-18，
  
  Published：16 October 2025
- 稿件说明：
移动端阅览
BI Guangjun, LONG Shixing, LUO Mingxing, et al. Partial Discharge Pattern Recognization Method for Generator Based on SURF-BoVF Features and Active Learning Model EQB-LightGBM[J]. High voltage apparatus, 2025, 61(10): 159-167.
DOI：

BI Guangjun, LONG Shixing, LUO Mingxing, et al. Partial Discharge Pattern Recognization Method for Generator Based on SURF-BoVF Features and Active Learning Model EQB-LightGBM[J]. High voltage apparatus, 2025, 61(10): 159-167. DOI： 10.13296/j.1001-1609.hva.2025.10.017.

摘要

为构建高效且高精度的发电机局部放电分类模型，提出了一种基于PRPD图谱SURF特征的视觉词袋表征方法以及主动学习EQB-LightGBM模式识别方法。首先，从5种发电机典型局部放电类型的PRPD图谱中提取SURF特征，这些特征被用于构建视觉词袋模型。通过该模型，发电机局部放电样本可以高效地量化为一个高维特征向量，从而有效地表征放电模式的特征组成。然后，将放电样本训练集划分为初始训练集和候选样本集，使用初始训练集训练分类投票模型，并通过EQB策略从候选样本集中筛选高价值样本，优化训练集参与模型迭代训练。通过动态更新训练集，提高了训练样本的代表性，从而提升了模型的分类识别性能。最后，在满足预设停止条件后，结束样本筛选与模型训练，得到高效且精准的识别模型。实验结果表明，文中所提方法筛选全训练集38%的样本进行训练即可达到98%的识别精度，与全训练集的识别精度相当，显著提高了模型识别效率，有效降低了训练时间和计算成本。

Abstract

For constructing an efficient and high-precision generator partial discharge classification model

a visual bag-of-words representation method based on SURF features of PRPD spectra

along with an active learning EQB-LightGBM pattern recognition approach is proposed. First

SURF features are extracted from the PRPD spectra of five typical partial discharge types

and these features are used to construct the visual bag-of-words model. With this model

partial discharge samples of generator are efficiently quantified into high-dimensional feature vectors

effectively representing the visual information of discharge patterns. Then

the discharge sample training set is divided into an initial training set and a candidate sample set. The former is used to train a classification voting model

and highvalue samples are selected from the latter through the EQB strategy to optimize the training set for iterative training of the model. By dynamically updating the training set

the representativeness of the training samples is improved

thereby improving the classification and recognition performance of the model. Finally

when the preset stopping conditions are met

the sample selection and model training processes are concluded and efficient and accurate recognition model is obtained. Experimental results show that the method proposed in this paper can achieve 98%recognition accuracy using only 38% of the training samples

which is comparable to the regognization accuracy achieved with the full training set. This method significantly improves model recognition efficiency and effectively reduces both training time and computational costs.

关键词

Keywords

references

李贵衡 , 张健 , 黄少坡 , 等 . 基于Wiener过程的电机主绝缘材料寿命预测方法 [J ] . 电机与控制学报 , 2023 , 27 ( 11 ): 40 - 47 .

LI Guiheng , ZHANG Jian , HUANG Shaopo , et al . Life prediction method of motor ground-wall insulation material based on Wiener process [J ] . Electric Machines and Control , 2023 , 27 ( 11 ): 40 - 47 .

高波 , 郝长金 , 尹小兵 , 等 . 发电机主绝缘加速老化的频域介电特性 [J ] . 电机与控制学报 , 2019 , 23 ( 3 ): 57 - 64 .

GAO Bo , HAO Changjin , YIN Xiaobing , et al . Frequency domain dielectric properties of accelerated aging generator stator insulation [J ] . Electric Machines and Control , 2019 , 23 ( 3 ): 57 - 64 .

ZHANG Shuqi , LI Chengrong , WANG Ke , et al . Improving recognition accuracy of partial discharge patterns by image-oriented feature extraction and selection technique [J ] . IEEE Transactions on Dielectrics and Electrical Insulation , 2016 , 23 ( 2 ): 1076 - 1087 .

CHEN Yuzhu , PENG Xiaosheng , WANG Hongyu , et al . Generator stator partial discharge pattern recognition based on PRPDGrabcut and DSC-GoogLeNet deep learning [J ] . IEEE Transactions on Dielectrics and Electrical Insulation , 2023 , 30 ( 5 ): 2267 - 2276 .

LI Ang , WEI Guangze , LI Shilong , et al . Pattern recognition of partial discharge in high-voltage cables using TFMT model [J ] . IEEE Transactions on Power Delivery , 2024 , 39 ( 6 ): 3326 - 3337 .

程江洲 , 温静怡 , 鲍刚 , 等 . 基于T-MobileNet-L模型的GIS局部放电模式识别研究 [J ] . 电子测量技术 , 2021 , 44 ( 20 ): 22 - 28 .

CHENG Jiangzhou , WEN Jingyi , BAO Gang , et al . Research on pattern recognition of GIS partial discharge based on T-MobileNet-L model [J ] . Electronic Measurement Technology , 2021 , 44 ( 20 ): 22 - 28 .

FIRUZIK , VAKILIAN M E , PHUNG B T , et al . Partial discharges pattern recognition of transformer defect model by LBP&HOG features [J ] . IEEE Trans on Power Delivery , 2019 , 34 ( 2 ): 542 - 550 .

汪可 , 廖瑞金 , 吴高林 , 等 . 采用双向改进模糊2DLDA算法提升多因素影响的局部放电识别可靠性 [J ] . 电工技术学报 , 2014 , 29 ( 11 ): 210 - 220 .

WANG Ke , LIAO Ruijin , WU Gaolin , et al . Improvement of partial discharge recognition reliability considering influence of multifactors based on two-directional modified fuzzy 2DLDA algorithm [J ] . Transactions of China Electrotechnical Society , 2014 , 29 ( 11 ): 210 - 220 .

汪可 , 张书琦 , 李金忠 , 等 . 基于灰度图像分解的局部放电特征提取与优化 [J ] . 电机与控制学报 , 2018 , 22 ( 5 ): 25 - 34 .

WANG Ke , ZHANG Shuqi , LI Jinzhong , et al . Partial discharge feature extraction and optimization based on gray image decomposition [J ] . Electric Machines and Control , 2018 , 22 ( 5 ): 25 - 34 .

KIM D , KIM M , KIM K , et al . Dynamic load balancing of parallel SURF with vertical partitioning [J ] . IEEE Transactions on Parallel and Distributed Systems , 2015 , 26 ( 12 ): 3358 - 3370 .

LI Yalin , ZHU Xinshan , LI Bin , et al . Discriminative features based comprehensive detector for defective insulators [J ] . Energy and AI , 2024 ( 17 ): 100387 .

李泽 , 王辉 , 钱勇 , 等 . 基于加速鲁棒特征的含噪局部放电模式识别 [J ] . 电工技术学报 , 2022 , 37 ( 3 ): 775 - 785 .

LI Ze , WANG Hui , QIAN Yong , et al . Pattern recognition of partial discharge in the presence of noise based on speeded up robust features [J ] . Transactions of China Electrotechnical Society , 2022 , 37 ( 3 ): 775 - 785 .

李航洋 , 阳同光 . 计及动态非均匀热特性的储能软包锂电池充放电热路模型 [J ] . 电力系统保护与控制 , 2023 , 51 ( 19 ): 104 - 113 .

LI Hangyang , YANG Tongguang . A charging/discharging thermal circuit model for energy storage pouch lithium batteries considering dynamic and non-uniform characteristics [J ] . Power System Protection and Control , 2023 , 51 ( 19 ): 104 - 113 .

GU Bin , ZHAI Zhou , DENG Cheng , et al . Efficient active learning by querying discriminative and representative samples and fully exploiting unlabeled data [J ] . IEEE Transactions on Neural Networks and Learning Systems , 2021 , 32 ( 9 ): 4111 - 4122 .

YU Hualong , YANG Xibei , ZHENG Shang , et al . Active learning from imbalanced data:A solution of online weighted extreme learning machine [J ] . IEEE Transactions on Neural Networks and Learning Systems , 2019 , 30 ( 4 ): 1088 - 1103 .

刘金海 , 赵真 , 付明芮 , 等 . 基于主动小样本学习的管道焊缝缺陷检测方法 [J ] . 仪器仪表学报 , 2022 , 43 ( 11 ): 252 - 261 .

LIU Jinhai , ZHAO Zhen , FU Mingrui , et al . Active small sample learning based the pipe weld defect detection method [J ] . Chinese Journal of Scientific Instrument , 2022 , 43 ( 11 ): 252 - 261 .

李恬 , 冯早 , 朱雪峰 . 基于主动学习和最优路径森林的管道故障分类识别方法 [J ] . 电子测量与仪器学报 , 2022 , 36 ( 12 ): 67 - 76 .

LI Tian , FENG Zao , ZHU Xuefeng . Pipeline blockage recognition method based on active learning and optimum-path forest [J ] . Journal of Electronic Measurement and Instrumentation , 2022 , 36 ( 12 ): 67 - 76 .

闵永智 , 郝大宇 , 王果 , 等 . 基于深度自适应K-means++算法的电抗器声纹聚类方法 [J ] . 电力系统保护与控制 , 2025 , 53 ( 8 ): 1 - 13 .

MIN Yongzhi , HAO Dayu , WANG Guo , et al . Reactor voiceprint clustering method based on deep adaptive K-means++ algorithm [J ] . Power System Protection and Control , 2025 , 53 ( 8 ): 1 - 13 .

PRADHAN M K , MINZ S , SHRIVASTAVA V K . Entropy query by bagging-based active learning approach in the extreme learning machine framework for hyperspectral image classification [J ] . Current Science , 2020 , 119 ( 6 ): 934 - 943 .

VERRELST J , DETHIERS , RIVERAJ P , et al . Active learning methods for efficient hybrid biophysical variable retrieval [J ] . IEEE Geoscience and Remote Sensing Letters , 2016 , 13 ( 7 ): 1012 - 1016 .

KARIMI M , MAJIDI M , MIRSAEEDI H , et al . A novel application of deep belief networks in learning partial discharge patterns for classifying corona, surface, and internal discharges [J ] . IEEE Transactions on Industrial Electronics , 2020 , 67 ( 4 ): 3277 - 3287 .

LI Weihua , HE Jingke , LIN Huibin , et al . A LightGBM-based multiscale weighted ensemble model for few-shot fault diagnosis [J ] , IEEE Transactions on Instrumentation and Measurement , 2023 ( 72 ): 1 - 14 .

Views

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

Distance-based Methods for Pattern Recogition of Discharge

STUDY OF FRACTAL GRID-DIMENSION PATTERN FOR PD RECOGNITION

APPLICATION OF FRACTALS THEORY TO PD PATTERN RECONGNITION FOR POWER EQUIPMENT

Recognition of Partial Discharge Distribution Patterns Based on Morlet Wavelet Transform and Neural Network

Partial Discharge Pattern Recognition of GIS Based on the UHF Signal Characteristics

Related Author

谈克雄

李福祺

LI Jian

SUN Caixin

CHEN Mingying

DU Lin

YUAN Zhijian

ZHAO Zhongyuan

Related Institution

清华大学

清华大学北京 100084

Chongqing University

重庆大学高电压与电工新技术教育部重点实?

Xi’an Jiaotong University

⁰