Power user classification and recognition method based on multimodal hybrid features

ZHANG Haijing; LIU Yijuan; SHAN Shuaijie; JIANG Yuan; FENG Yankun

doi:10.11930/j.issn.1004-9649.202504090

English Version Journal Login

您当前的位置：

首页 >

文章列表页 >

Power user classification and recognition method based on multimodal hybrid features

更新时间：2026-07-01

- Power user classification and recognition method based on multimodal hybrid features
- Vol. 59, Issue 6, Pages: 112-124(2026)
- 作者机构：
  
  1. 国网山东省电力公司,山东,济南,250000
  2. 山东大学电气工程学院,山东,济南,250061
- 作者简介：
- 基金信息：
- DOI：10.11930/j.issn.1004-9649.202504090
  CLC：
- Online First：22 June 2026，
  
  Published：2026
- 稿件说明：
移动端阅览
张海静, 刘昳娟, 单帅杰, 江源, 冯延坤. 基于多模态混合特征的电力用户分类识别方法[J]. 中国电力, 2026, 59(6): 112-124.

ZHANG Haijing, LIU Yijuan, SHAN Shuaijie, et al. Power user classification and recognition method based on multimodal hybrid features[J]. 2026, 59(6): 112-124.
张海静, 刘昳娟, 单帅杰, 江源, 冯延坤. 基于多模态混合特征的电力用户分类识别方法[J]. 中国电力, 2026, 59(6): 112-124. DOI： 10.11930/j.issn.1004-9649.202504090.

ZHANG Haijing, LIU Yijuan, SHAN Shuaijie, et al. Power user classification and recognition method based on multimodal hybrid features[J]. 2026, 59(6): 112-124. DOI： 10.11930/j.issn.1004-9649.202504090.

摘要

针对代理购电用户用电行为的不规则性和多样性，为实现电力用户类型精准识别，提出了一种基于一维统计特征、格拉姆角场和递归图的多模态混合特征的分类识别方法。首先，根据现行分时电价政策，将用户年度用电数据划分为春、夏、秋、冬4个典型用电曲线，并针对工作日和节假日用电行为的差异性，提取用户节假日典型用电曲线。然后，基于这5类典型用电曲线，计算用户的电价敏感系数、用电平稳度等统计特征，并结合形状距离，利用K-shapes聚类方法生成用户类型标签。最后，通过格拉姆角场和递归图将用户年分时用电曲线进行图像化处理，与统计特征相结合，构建多模态混合特征的用户分类识别模型。实验结果表明，基于形状距离的K-shapes聚类方法能够精准划分用户类型，所提出的用户类型识别方案对各聚类标签用户的识别准确率均超过94%，可以为代理购电公司的用户侧管理提供有效的技术支撑。

Abstract

To address the irregularity and diversity of electricity consumption behavior among proxy power purchase users

this paper proposes a classification and recognition method based on multimodal hybrid features combining one-dimensional statistical features

Gramian Angular Field (GAF)

and Recurrence Plot (RP)

namely

1D-GAFs-RP

to realize accurate identification of power consumer types. Initially

in accordance with the current time-of-use electricity price policy

the annual electricity consumption data of users are divided into four typical consumption curves corresponding to spring

summer

autumn

and winter. Additionally

considering the differences in electricity consumption behaviors between weekdays and holidays

the typical consumption curves for holidays are extracted. Subsequently

based on these five typical consumption curves

the statistical features such as users' price sensitivity coefficient and electricity consumption stability are calculated. Combining these with shape distances

the k-shapes clustering method is employed to generate user type labels. Finally

by applying GAFs and RPs to visualize the annual time-of-use electricity consumption curves and integrating them with statistical features

a multimodal hybrid feature-based user classification and recognition model is constructed. Experimental results indicate that the k-shapes clustering method based on shape distances can accurately classify user types. The proposed user type recognition scheme achieves an identification accuracy rate of over 94% for each cluster label

providing effective technical support for user-side management of proxy power purchase companies.

关键词

Keywords

references

1宋军英, 崔益伟, 李欣然, 等. 基于欧氏动态时间弯曲距离与熵权法的负荷曲线聚类方法[J]. 电力系统自动化, 2020, 44 (15): 87- 98.

SONG Junying, CUI Yiwei, LI Xinran, et al. Load curve clustering method based on euclidean dynamic time warping distance and entropy weight[J]. Automation of Electric Power Systems, 2020, 44 (15): 87- 98.

2钱科军, 沈杰, 刘乙, 等. 基于负荷聚类的居民需求响应积分精准激励机制[J]. 智慧电力, 2019, 47 (7): 29- 35.

QIAN Kejun, SHEN Jie, LIU Yi, et al. Accurate score incentive mechanism of resident demand response based on load clustering[J]. Smart Power, 2019, 47 (7): 29- 35.

3徐邦恩, 蔺红. 基于改进模糊聚类的典型日负荷曲线选取方法[J]. 电测与仪表, 2019, 56 (4): 21- 26.

XU Bang'en, LIN Hong. Selection method of typical daily load curve based on improved fuzzy clustering[J]. Electrical Measurement & Instrumentation, 2019, 56 (4): 21- 26.

4金伟超, 张旭, 刘晟源, 等. 基于剪枝策略和密度峰值聚类的行业典型负荷曲线辨识[J]. 电力系统自动化, 2021, 45 (4): 20- 28.

JIN Weichao, ZHANG Xu, LIU Shengyuan, et al. Identification of typical industrial power load curves based on pruning strategy and density peak clustering[J]. Automation of Electric Power Systems, 2021, 45 (4): 20- 28.

5董哲, 陈玉梁, 薛同来, 等. 基于全局与滑动窗口结合的Attention机制的非侵入式负荷分解算法[J]. 电测与仪表, 2023, 60 (11): 74- 80.

DONG Zhe, CHEN Yuliang, XUE Tonglai, et al. Non-intrusive load monitoring algorithm based on Attention mechanism combined with global and sliding window[J]. Electrical Measurement & Instrumentation, 2023, 60 (11): 74- 80.

6严宇, 孙浩泳, 柏兆勇, 等. 基于MEFE和GN-CNN的低压台区用户用电行为分析方法[J]. 供用电, 2023, 40 (12): 79- 86.

YAN Yu, SUN Haoyong, BAI Zhaoyong, et al. Analysis method of user's electricity consumption behavior in low-voltage station area based on MEFE and GN-CNN[J]. Distribution & Utilization, 2023, 40 (12): 79- 86.

7吴昀烔, 赵健, 宣羿, 等. 基于多维负荷特性挖掘的电力特殊用户用电行为分析[J]. 电力建设, 2024, 45 (3): 116- 125.

WU Yuntong, ZHAO Jian, XUAN Yi, et al. Analysis of power consumption behavior of special users based on multidimensional load characteristic mining[J]. Electric Power Construction, 2024, 45 (3): 116- 125.

8李丹, 贺帅, 颜伟, 等. 考虑动态时间锚点和典型特征约束的年日均负荷曲线预测[J]. 中国电力, 2024, 57 (11): 36- 47.

LI Dan, HE Shuai, YAN Wei, et al. Annual daily average load curve prediction considering dynamic time anchors and typical feature constraints[J]. Electric Power, 2024, 57 (11): 36- 47.

9NOVIANTI P, SETYORINI D, RAFFLESIA U. K-Means cluster analysis in earthquake epicenter clustering[J]. International Journal of Advances in Intelligent Informatics, 2017, 3 (2): 81.

10KONG Q K, TRUGMAN D T, ROSS Z E, et al. Machine learning in seismology: turning data into insights[J]. Seismological Research Letters, 2019, 90 (1): 3- 14.

11PANAPAKIDIS I P, CHRISTOFORIDIS G C. Implementation of modified versions of the K-means algorithm in power load curves profiling[J]. Sustainable Cities and Society, 2017, 35, 83- 93.

12PANAPAKIDIS I, ASIMOPOULOS N, DAGOUMAS A, et al. An improved fuzzy C-means algorithm for the implementation of demand side management measures[J]. Energies, 2017, 10 (9): 1407.

13崔高颖, 邵雪松, 陈霄, 等. 基于多维特征分析与动态定权聚类的电力居民用户分类[J]. 电力需求侧管理, 2023, 25 (6): 88- 94.

CUI Gaoying, SHAO Xuesong, CHEN Xiao, et al. Residential electricity users classification based on multidimensional feature analysis and dynamic weighted clustering[J]. Power Demand Side Management, 2023, 25 (6): 88- 94.

14CEN X, CHEN Z L, CHEN H F, et al. User repurchase behavior prediction for integrated energy supply stations based on the user profiling method[J]. Energy, 2024, 286, 129625.

15陆俊, 朱炎平, 彭文昊, 等. 智能用电用户行为分析特征优选策略[J]. 电力系统自动化, 2017, 41 (5): 58- 63, 83.

LU Jun, ZHU Yanping, PENG Wenhao, et al. Feature selection strategy for electricity consumption behavior analysis in smart grid[J]. Automation of Electric Power Systems, 2017, 41 (5): 58- 63, 83.

16周琪, 杨洁, 韩俊杰, 等. 基于大数据的业扩用户用电行为特征研究[J]. 中国电力, 2017, 50 (10): 176- 180.

ZHOU Qi, YANG Jie, HAN Junjie, et al. Research on electricity consumption behaviors of the capacity expansion users based on big data analysis[J]. Electric Power, 2017, 50 (10): 176- 180.

17沈建良, 陆春光, 袁健, 等. 大数据技术在电力大用户用电特征分类中的应用[J]. 浙江电力, 2017, 36 (12): 37- 41.

SHEN Jianliang, LU Chunguang, YUAN Jian, et al. Application of big data technology in power consumption feature classification of large customers[J]. Zhejiang Electric Power, 2017, 36 (12): 37- 41.

18杨明, 李梦林, 王勃, 等. 面向新型电力系统的数值天气预报技术及应用综述[J]. 电力系统自动化, 2025, 49 (17): 1- 20.

YANG Ming, LI Menglin, WANG Bo, et al. Review on technologies and applications of numerical weather prediction for new power system[J]. Automation of Electric Power Systems, 2025, 49 (17): 1- 20.

19王宇飞, 杜桐, 边伟国, 等. 基于DTW K-medoids与VMD-多分支神经网络的多用户短期负荷预测[J]. 中国电力, 2024, 57 (6): 121- 130.

WANG Yufei, DU Tong, BIAN Weiguo, et al. Short-term load forecasting based on DTW K-medoids and VMD multi-branch neural network for multiple users[J]. Electric Power, 2024, 57 (6): 121- 130.

20ZHU J, MIAO Y, DONG H, et al. Short-term residential load forecasting based on K-shape clustering and domain adversarial transfer network[J]. Journal of Modern Power Systems and Clean Energy, 2024, 12 (4): 1239- 1249.

21MARTÍN A G, FERNÁNDEZ-ISABEL A, DIEGO I M D, et al. A survey for user behavior analysis based on machine learning techniques: current models and applications[J]. Applied Intelligence, 2021, 51 (8): 6029- 6055.

22SHAN S J, LIU J B, WU S G, et al. A motor bearing fault voiceprint recognition method based on Mel-CNN model[J]. Measurement, 2023, 207, 112408.

23吴郅君, 殷新博, 陈中, 等. 基于模糊聚类曲线相似度的负荷用户识别方法[J]. 电力工程技术, 2019, 38 (3): 151- 156.

WU Zhijun, YIN Xinbo, CHEN Zhong, et al. Identification method of load customers based on similarity of fuzzy clustering curves[J]. Electric Power Engineering Technology, 2019, 38 (3): 151- 156.

24LEI X H, XIE F D, JIN C. Gram angle field-based Siamese graph convolutional neural network for hyperspectral images classification[J]. IEEE Geoscience and Remote Sensing Letters, 2025, 22, 5501305.

25丁伟, 宋俊材, 陆思良, 等. 基于多通道信号二维递归融合和ECA-ConvNeXt的永磁同步电机高阻接触故障诊断[J]. 电工技术学报, 2024, 39 (20): 6397- 6408.

DING Wei, SONG Juncai, LU Siliang, et al. High-resistance connection fault diagnosis of permanent magnet synchronous motor based on two-dimensional recursive fusion of multi-channel signals and ECA-ConvNeXt[J]. Transactions of China Electrotechnical Society, 2024, 39 (20): 6397- 6408.

26YANG Z J, DIAO C Y, LI B. A robust hybrid deep learning model for spatiotemporal image fusion[J]. Remote Sensing, 2021, 13 (24): 5005.

27李欣, 付豫韬, 李新宇, 等. 基于GAF-CNN的电力系统暂态稳定评估[J]. 智慧电力, 2023, 51 (11): 45- 52.

LI Xin, FU Yutao, LI Xinyu, et al. Power system transient stability assessment based on GAF-CNN[J]. Smart Power, 2023, 51 (11): 45- 52.

28PAPARRIZOS J, GRAVANO L. K-shape: efficient and accurate clustering of time series[J]. ACM SIGMOD Record, 2015, 45(1): 69–76.

29张晓珂, 张辉, 戴小然, 等. 基于层次聚类和BILSTM的光伏短期功率预测模型[J]. 智慧电力, 2024, 52 (9): 41- 48.

ZHANG Xiaoke, ZHANG Hui, DAI Xiaoran, et al. Photovoltaic short-term power forecasting model based on hierarchical clustering & BILSTM[J]. Smart Power, 2024, 52 (9): 41- 48.

30窦翔, 李卓群, 张哲, 等. 基于CNN-BiLSTM-RF-KDE的综合能源系统负荷预测[J]. 综合智慧能源, 2025, 47 (9): 60- 70.

DOU Xiang, LI Zhuoqun, ZHANG Zhe, et al. Load forecasting for integrated energy systems based on CNN-BiLSTM-RF-KDE[J]. Integrated Intelligent Energy, 2025, 47 (9): 60- 70.

31赵贵中, 黄淼华. 基于BERT-BiLSTM-CRF的电力事故信息抽取方法[J]. 综合智慧能源, 2024, 46 (11): 19- 28.

ZHAO Guizhong, HUANG Miaohua. An information extraction method for electric power accidents based on BERT-BiLSTM-CRF model[J]. Integrated Intelligent Energy, 2024, 46 (11): 19- 28.

Views

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

Analysis on the spatiotemporal evolution mechanism of electricity consumption inequality in China

Deep reinforcement learning-driven decision-making paradigm for electricity-carbon-hydrogen collaboration

Wind power prediction based on PCA and SOFTS fusion model

Equivalent modeling of large-scale photovoltaic hydrogen production stations driven by physics-data collaboration

Equivalent energy storage model for district heating network and its application in non-iterative combined electricity-heat scheduling

Related Author

No data

Related Institution

No data

AI问答

⁰