作者：严勤¹, 邓高峰², 胡涛², 胡志强¹, 马建²

作者单位：1. 国网江西省电力有限公司，江西 南昌 330096;
2. 国网江西省电力有限公司供电服务管理中心，江西 南昌 330096

关键词：高级计量架构;窃电检测;深度循环网络;门控循环单元;长短时记忆网络

摘要：

现有的窃电检测方法通常利用电力用户的静态特征和浅层的检测模型，没有充分利用隐含在数据下的时序特征。为此，该文提出基于双向深度循环神经网络的窃电检测方法，分别采用门控循环单元和长短时记忆网络建立双向深度循环神经网络模型，输入用户的用电量数据，利用循环神经网络提取数据的时序特性，将时序特征输入反向传播神经网络进行分类。对爱尔兰社会科学数据档案馆提供的电力用户行为试验数据进行实验分析，该数据集包含5000个家庭和企业用户超过一年的用电量数据，采样时间为30 min。结果表明，与传统浅层的神经网络模型相比，双向深度循环神经网络的窃电检测方法具有更高的准确度和鲁棒性。

Abnormal electricity detection method based on deep recurrent neural network
YAN Qin¹, DENG Gaofeng², HU Tao², HU Zhiqiang¹, MA Jian²
1. State Grid Jiangxi Electric Power Co., Ltd., Nanchang 330096, China;
2. Power Supply Service Management Center of State Grid Jiangxi Electric Power Co., Ltd., Nanchang 330096, China
Abstract: Existing electricity theft detection methods usually use the static characteristics of power users and shallow detection models, and do not make full use of the timing characteristics hidden in the data. To this end, this paper proposes an electricity theft detection method based on a bidirectional deep recurrent neural network, which uses a gated recurrent unit and a long-short-term memory network to establish a bidirectional deep recurrent neural network model, input user power consumption data, and use the recurrent neural network to extract data time series characteristics, input the time series characteristics into the back propagation neural network for classification. Conducted an experimental analysis of the power user behavior test data provided by the Irish Social Science Data Archive. The data set contains electricity consumption data for

households and business users for more than one year, and the sampling time is 30 minutes. The results show that, compared with the traditional shallow neural network model, the bidirectional deep recurrent neural network detection method has higher accuracy and robustness.
Keywords: advanced metering infrastructure;electricity theft detection;deep recurrent neural network;gated recurrent unit;long-short-term memory network
2021, 47(7):99-104  收稿日期: 2020-06-02;收到修改稿日期: 2020-07-02
基金项目: 国家电网科技项目资助（52182019000H）
作者简介: 严勤(1982-),男,江西新余市人,高级工程师,研究方向为计量技术研究与管理
参考文献
[1] 梁跃. 基于机器学习的电力用户用电异常检测技术研究[D]. 北京: 北京邮电大学, 2019.
[2] SHEN R J R L Y W J L C S X. Energy-theft detection issues for advanced metering infrastructure in smart grid[J]. Tsinghua Science and Technology, 2014, 19(2): 105-120
[3] MCLAUGHLIN P M S. Security and privacy challenges in the smart grid[J]. IEEE Security & Privacy, 2009, 7(3): 75-77
[4] XIA X, XIAO Y, LIANG W. SAI: A suspicion assessment-based inspection algorithm to detect malicious users in smart grid[J]. Ieee Transactions on Information Forensics and Security, 2020, 15: 361-374
[5] AHMAD T, CHEN H, WANG J, et al. Review of various modeling techniques for the detection of electricity theft in smart grid environment[J]. Renewable & Sustainable Energy Reviews, 2018, 82: 2916-2933
[6] 李博. 基于深度学习的用户窃电行为检测[D]. 北京: 北京邮电大学, 2019.
[7] KHAZAEE A, SAFA H H, GHASEMPOUR M, et al. Distribution loss reduction in residential and commercial pilots by using AMI system[J]. CIRED - Open Access Proceedings Journal, 2017, 2017(1): 1711-1714
[8] 叶興, 薛家祥. 改进型LSTM网络光伏发电功率预测研究[J]. 中国测试, 2019, 45(11): 14-20
[9] 刘帅, 刘长良, 曾华清. 基于核极限学习机的风电机组齿轮箱故障预警研究[J]. 中国测试, 2019, 45(2): 121-127
[10] 杨婧, 辛明勇, 欧家祥, 等. 基于大数据的配电网线损定位与评估方法研究[J]. 中国测试, 2019, 45(7): 19-24
[11] JAIN S, CHOKSI K A, PINDORIYA N M. Rule-based classification of energy theft and anomalies in consumers load demand profile[J]. IET Smart Grid, 2019, 2(4): 612-624
[12] PUNMIYA R, CHOE S. Energy theft detection using gradient boosting theft detector with feature engineering-based preprocessing[J]. IEEE Transactions on Smart Grid, 2019, 10(2): 2326-2329
[13] ZHENG Z, YANG Y, NIU X, et al. Wide and deep convolutional neural networks for electricity-theft detection to secure smart grids[J]. IEEE Transactions on Industrial Informatics, 2018, 14(4): 1606-1715
[14] Irish Social Science data archive.[EB/OL].[2020-05-28]. http://www.ucd.ie/issda/data/commissionforenergyregulationcer/.
[15] JOKAR P, ARIANPOO N, LEUNG V C M. Electricity theft detection in AMI using customers' consumption patterns[J]. IEEE Transactions on Smart Grid, 2016, 7(1): 216-226
[16] HE H B, BAI Y, GARCIA E A, et al. ADASYN: Adaptive synthetic sampling approach for imbalanced learning[J]. IEEE IJCNN, 2008: 1322-1328
[17] SCHMIDHUBER S H J. Long short-term memory[J]. Neural Computation, 1997, 9(8): 1735-1780