Identifying electricity theft based on residual network and depthwise separable convolution enhanced self attention

doi:10.3969/j.issn.1000-5641.2023.05.016

Abstract

Abstract:

Power theft seriously endangers power equipment and personal safety, and causes significant economic losses for energy suppliers. Hence, it is important for these suppliers to accurately identify instances of power theft to reduce losses and increase efficiency. In this paper, based on the residual network (ResNet) structure, a 2D convolutional neural network is combined with a depthwise separable convolution enhanced self-attentive (DSCAttention) mechanism to improve the number of correctly-classified electricity theft users. In addition, electricity theft data often contains missing values, outliers, and positive and negative sample imbalance. Each of the above problems are treated separately using the zero-completion method, quantile transformation, and hierarchical splitting method, respectively. The proposed model has been extensively tested using real power theft data sets. The results show that the area under curve (AUC) index of the proposed model reaches a value of 91.92%, while mean average precision values MAP@100 and MAP@200 are measured reaching 98.58% and 96.77%, respectively. Compared with other electricity theft classification models, the proposed model performs the electricity theft classification task better. The method in this paper can be extended to electricity theft intelligent identification.

Key words: residual network, convolution enhancement, self-attention mechanism, depthwise separable convolution, identifying electricity theft

CLC Number:

TP393

Zhishang DUAN, Yi RAN, Duliang LYU, Jie QI, Jiachen ZHONG, Peisen YUAN. Identifying electricity theft based on residual network and depthwise separable convolution enhanced self attention[J]. Journal of East China Normal University(Natural Science), 2023, 2023(5): 193-204.

Figures/Tables 14

Fig.1

Fig.2

Fig.3

Table 1

Fig.4

Fig.5

Fig.6

Table 2

Fig.7

Fig.8

Fig.9

Fig.10

Table 3

Fig.11

References 23

1	XIA R, GAO Y P, ZHU Y Q, et al.. An attention-based wide and deep CNN with dilated convolutions for detecting electricity theft considering imbalanced data. Electric Power Systems Research, 2023, 214 (A): 108886.
2	GLAUNER P, MEIRA J A, VALTCHEV P, et al.. The challenge of non-technical loss detection using artificial intelligence: A survey. International Journal of Computational Intelligence Systems, 2017, 10 (1): 760- 775.
3	BUZAU M M, TEJEDOR-AGUILERA J, CRUZ-ROMERO P, et al.. Detection of non-technical losses using smart meter data and supervised learning. IEEE Transactions on Smart Grid, 2018, 10 (3): 2661- 2670.
4	XIA X F, XIAO Y, LIANG W, et al.. Detection methods in smart meters for electricity thefts: A survey. Proceedings of the IEEE, 2022, 110 (2): 273- 319.
5	YATSKO V A.. Patterns of using the Z-score for text classification purposes . Automatic Documentation and Mathematical Linguistics, 2022, 56 (5): 245- 250.
6	周赣, 华济民, 李铭钧, 等.. 基于图转换和混合卷积神经网络的窃电检测方法. 电力系统自动化, 2022, 46 (19): 78- 86.
7	LI S, HAN Y H, YAO Y, et al. Electricity theft detection in power grids with deep learning and random forests[J]. Journal of Electrical and Computer Engineering, 2019: Article ID 4136874. DOI: 10.1155/2019/4136874.
8	李强, 张立梅, 白牧可.. 基于多元数据特征和改进随机森林的智能配电网异常数据辨识. 科学技术与工程, 2023, 23 (5): 2007- 2015.
9	ROUZBAHANI H M, KARIMIPOUR H, LEI L. An ensemble deep convolutional neural network model for electricity theft detection in smart grids[C]// 2020 IEEE International Conference on Systems, Man, and Cybernetics (SMC). IEEE, 2020: 3637-3642.
10	徐扬, 张耀, 陈宇轩, 等.. 基于Bagging混合策略的多风电场稀疏向量自回归概率预测. 电力系统保护与控制, 2023, 51 (7): 95- 106.
11	谭庆康, 潘沛生.. 基于注意力机制的堆叠LSTM心电预测算法. 计算机技术与发展, 2023, 33 (1): 62- 67.
12	赵星宇, 吴泉军, 朱威.. 基于CEEMDAN和TCN-LSTM模型的短期电力负荷预测. 科学技术与工程, 2023, 23 (4): 1557- 1564.
13	CHEN W, SHI K.. Multi-scale attention convolutional neural network for time series classification. Neural Networks, 2021, 136 (2): 126- 140.
14	HASAN M N, TAMA R N, NAHID A A, et al.. Electricity theft detection in smart grid systems: A CNN-LSTM based approach. Energies, 2019, 12 (17): 3310- 3328.
15	简定辉, 李萍, 黄宇航.. 基于GA-VMD-ResNet-LSTM网络的短期电力负荷预测. 国外电子测量技术, 2022, 41 (10): 15- 22.
16	LI Y X, CAO W C, DROSSOS K, et al. Domestic activity clustering from audio via depthwise separable convolutional autoencoder network[C]// 2022 IEEE 24th International Workshop on Multimedia Signal Processing (MMSP). IEEE, 2022. DOI: 10.1109/MMSP55362.2022.9949512.
17	WU H X, HU T G, LIU Y, et al. TimesNet: Temporal 2D-variation modeling for general time series analysis [EB/OL]. (2023-04-12)[2023-05-11]. https://doi.org/10.48550/arXiv.2210.02186.
18	ZHENG Z B, YANG Y T, NIU X D, et al.. Wide and deep convolutional neural networks for electricity-theft detection to secure smart grids. IEEE Transactions on Industrial Informatics, 2018, 14 (4): 1606- 1615.
19	WANG Z G, OATES T. Imaging time-series to improve classification and imputation[C]// Proceedings of the 24th International Conference on Artificial Intelligence (IJCAI’15). AAAI, 2015: 3939-3945.
20	赵晋斌, 张建平, 毛玲, 等.. 基于PSO-Soft attention双向LSTM算法的光伏发电量预测研究. 智慧电力, 2022, 50 (3): 1- 7.
21	皇甫晓瑛, 钱惠敏, 黄敏.. 结合注意力机制的深度神经网络综述. 计算机与现代化, 2023, (2): 40- 49.
22	VASWANI A, SHAZEER N, PARMAR N, et al. Attention is all you need[C]// Proceedings of the 31st International Conference on Neural Information Processing Systems (NIPS’17). Red Hook, NY, USA: Curran Associates Inc., 2017: 6000-6010.
23	HEARST M A, DUMAIS S T, OSUNA E, et al.. Support vector machines. IEEE Intelligent Systems and Their Applications, 1998, 13 (4): 18- 28.

ResNet18参数	本文ResNet结构参数
Conv7-64, 步长2	Conv3-8, 步长1
maxpool, 步长2	maxpool, 步长(2,1)
Conv3-64	Conv3-16 Conv3-16 depthwise3-16	attention-8 depthwise3-8
Conv3-64
Conv3-128	pointwise1-48
Conv3-128
Conv3-256	FC-1024
Conv3-256
Conv3-512	FC-2
Conv3-512
averagepool	softmax
FC-1000	–
softmax	–

样本描述	数据值
每条样本时间	2014-01-01—2016-10-31
正常用电户样本	38757 条
窃电用户样本	3615 条
存在缺失值样本占比	25%

模型	AUC	MAP@100	MAP@200
RF	0.8008	0.8425	0.7615
SVM	0.8158	0.8935	0.8373
WDCNN	0.8046	0.6757	0.6701
本文模型	0.9192	0.9858	0.9677

[1]	Yang ZHANG, Yejing LAI, Dingjiang HUANG. Device component state recognition method of power distribution cabinet based on a residual networks [J]. Journal of East China Normal University(Natural Science), 2023, 2023(2): 132-142.
[2]	Caidie HUANG, Xinping WANG, Liangyu CHEN, Yong LIU. Research on a knowledge tracking model based on the stacked gated recurrent unit residual network [J]. Journal of East China Normal University(Natural Science), 2022, 2022(6): 68-78.
[3]	FU Yu, LI You, LIN Yu-ming, ZHOU Ya. Self-attention based neural networks for product titles compression [J]. Journal of East China Normal University(Natural Sc, 2019, 2019(5): 113-122,167.