Abnormal Power Consumption Mode Detection Based on Empirical Mode Decomposition and Multi-View Clustering

doi:10.16628/j.cnki.2095-8188.2023.03.012

Abstract

Abstract:

In order to solve the low detection efficiency of the existing abnormal power consumption detection methods,the anomaly detection method based on empirical mode and multi view clustering is proposed.Following the process of "empirical mode decomposition-dimensional constraints-multi-view clustering-horizontal detection-vertical detection" and combining the multi-view clustering with the preliminary criteria,the detection rate is significantly improved.In the anomaly detection algorithm,the grid-based entropy outlier factor (Grid-EOF) algorithm is proposed.A new criterion is given based on the longitudinal detection,which can improve the detection rate of users with unknown electricity theft.Finally,it is verified by the measured data of smart meters of the State Grid of China.The results show that the introduction of multi-view clustering,improved algorithm and longitudinal detection can effectively improve the detection rate and accuracy of the anomaly detection model.

Key words: abnormal electricity utilization detection, empirical mode decomposition, multi-view clustering, Shannon entropy

CLC Number:

TM930

WANG Jianyuan, LIU Kechen. Abnormal Power Consumption Mode Detection Based on Empirical Mode Decomposition and Multi-View Clustering[J]. LOW VOLTAGE APPARATUS, 2023, 0(3): 73-80.

Figures/Tables 9

References 26

[1]	魏梅芳, 甘明, 王宇. 基于用电信息采集系统的反窃降损分析[J]. 湖南电力, 2016, 36(6):48-50.
[2]	杨佳. 基于电网电气参量特性分析的窃电辨识方法[D]. 保定: 华北电力大学, 2013.
[3]	LIU S X, YING L, WANG J L, et al. Identification of stealing electricity based on big data analysis[J]. Energy Reports, 2020, 6(9):731-738.
[4]	CHENG C, ZHANG H J, JING Z M, et al. Study on the anti-electricity stealing based on outlier algorithm and the electricity information acquisition system[J]. Power System Protection Control, 2015, 43(17):69-73.
[5]	朱天怡, 艾芊, 李昭昱, 等. 一种数据驱动的用电行为分析模型研究[J]. 电器与能效管理技术, 2019(19):91-100.
[6]	LIU J, CAO X, WANG D G, et al. Abnormal electricity detection with hybrid deep neural network model[J]. MATEC Web of Conferences, 2018, 189(7):03001. doi: 10.1051/matecconf/201818903001
[7]	LIAO W L, YANG Z, LIU K P, et al. Electricity theft detection using euclidean and graph convolutional neural networks[J]. IEEE Transactions on Power Systems, 2022, 13(2):1-13.
[8]	吴少雷, 肖坚红, 张波, 等. 基于阻抗分析的居民用户异常用电研判模型研究[J]. 电器与能效管理技术, 2019(3):30-34,67.
[9]	刘伟, 马世乾, 李云飞, 等. 多源数据驱动的配电网网损精确分析方法[J]. 电力系统及其自动化学报, 2022, 34(11):12-16.
[10]	庞传军, 余建明, 冯长有, 等. 基于LSTM自动编码器的电力负荷聚类建模及特性分析[J]. 电力系统自动化, 2020, 44(23):57-63.
[11]	李子康, 刘灏, 毕天姝, 等. 基于数据驱动的鲁棒性电力系统扰动识别[J]. 中国电机工程学报, 2021, 41(21):7261-7274.
[12]	袁非牛, 章琳, 史劲亭, 等. 自编码神经网络理论及应用综述[J]. 计算机学报, 2019, 42(1):203-230.
[13]	庄池杰, 张斌, 胡军, 等. 基于无监督学习的电力用户异常用电模式[J]. 中国电机工程学报, 2016, 36(2):379-387.
[14]	孙毅, 李世豪, 崔灿, 等. 基于高斯核函数改进的电力用户用电数据离群点检测方法[J]. 电网技术, 2018, 42(3):1595-1604.
[15]	李宁, 尹小明, 丁学峰, 等. 一种融合聚类和异常检测算法的窃电辨识方法[J]. 电测与仪表, 2018, 55(21):19-24.
[16]	GUO L Q, CHEN X B, LIU Z M, et al. Detection method for power theft based on SOM neural network and k-means clustering algorithm[C]// International Conference on Electrical Machines and Systems, 2019:1-5.
[17]	宋易阳, 李存斌, 祁之强, 等. 基于云模型和模糊聚类的电力负荷模式提取方法[J]. 电网技术, 2014, 38(12):3378-3383.
[18]	王浩. 多视角数据聚类研究[D]. 成都: 西南交通大学, 2020.
[19]	ZHANG P J, WANG C. A type Ⅲ radio burst[J]. Astronomy & Astroph, 2018, 618:A165.
[20]	HE Z Y, XU X F, DENG S C. Discovering cluster-based local outliers[J]. Pattern Recognition Letters, 2003, 24(9):1641-1650. doi: 10.1016/S0167-8655(03)00003-5
[21]	HUANG N E, SHEN Z, LONG S R, et al. The empirical mode decomposition and the hilbert spectrum for nonlinear and non-stationary time series analysis[J]. Proceeding of Rotal Society of London:Series A,1998, 454(1971):903-995.
[22]	WANG H, YANG Y. GMC:Graph-Based Multi-View Clustering[J]. IEEE Transactions on knowledge and data engineering, 2020, 32(6):1116-1128. doi: 10.1109/TKDE.69
[23]	杨茂林. 离群检测算法研究[D]. 武汉: 华中科技大学, 2012.
[24]	STEINARSSON S. Downsampling time series for visual represention[D]. Iceland: University of Iceland, 2013.
[25]	ZHENG Z B, NIU X D, DAI H N, et al. Wide and deep convolutional neural networks for electricity-theft detection to secure smart grids[J]. IEEE Transsactions on Industrial Informatics, 2018, 14(4):1606-1615.
[26]	王奉涛, 苏文盛. 滚动轴承故障诊断与寿命预测[M]. 北京: 科学出版社, 2018.

类别号	总数/异常用户数
1	1 591/30
2	1 460/21
3	1 357/5
4	48/48

模型	TPR/%	FPR/%	AUC
K-means+时域指标+ LOF异常检测	50.96	5.99	0.840 2
模糊C均值聚类+时域指标+LOF异常检测	53.84	6.16	0.851 1
(IMF1+INSA)+GMC +Grid-EOF异常检测	78.85	2.96	0.920 3
(IMF1+INSA)+GMC+ Grid-EOF异常检测+纵向检测	89.42	2.81	0.950 5