随着经济的快速增长和科学技术水平的飞速发展，居民对电力的需求不断增长，电网容量不断扩大，智能电网发展速度也与日俱增。然而，用户的异常用电行为越来越频繁，及时有效地检测电力系统中的异常用电行为对于发现恶意的异常用电用户、减少经济损失，从而确保电力系统安全可靠地运行、维护国民经济平稳发展至关重要。随着高级计量基础设施的部署，积累了海量的大规模高维复杂电力能耗数据，但存在数据缺失和异常用电检测精度低等问题。如何对电力能耗数据的缺失值进行有效填补以提高数据质量，并从数据层面对异常用电用户进行有效的检测，近年已受到广泛关注。论文研究的目的是提高电力能耗数据缺失值的填补准确度和异常用电行为检测的精度。

考虑到高维电力能耗数据邻近特征之间的相关性以及潜在的时空特征，本文利用深度学习，对电力能耗数据进行了分析，提高了异常用电检测精度，主要工作内容如下：

（1）针对现有的对时序数据的缺失值填补模型中，未考虑到邻近数据特征相关性，且缺失数据拟合值不符合真实数据分布规律的问题，在缺失值填补算法的基础上引入多尺度滑动窗口，自适应地提取时序数据的特征，捕捉邻近数据中潜在的相关性，提出一种多尺度滑动窗口的生成对抗网络模型（MSW-GAN）进行缺失值填补。在MSW-GAN的生成器输入中引入提示矩阵以加快模型训练，抑制判别器能力。在多个包含时序特性的数据上进行了实验，验证了提出的模型对电力能耗数据的缺失值填补更精确。

（2）针对现有的电力盗窃检测模型中，未考虑到电力能耗数据集的时序特征和空间特征，对提取的特征表征能力不足的问题，提出了一种混合卷积自编码器和长短期记忆网络的深度学习模型（CAEs-LSTM）检测有异常用电行为的用户。通过长短期记忆网络学习电力能耗数据全局的时序特征，卷积自编码器学习空间特征，并进行高斯白噪声处理以提高模型抗噪能力。通过由国家电网公布的真实数据集进行实验，结果表明混合模型有效提高了检测精度。

With the rapid growth of the economy and the rapid development of scientific and technical, residents have grown to electricity, and the power grids capacity is expanding. The development speed of power grids is also increasing. However, in recent years, the user's abnormal electrical behavior is increasingly frequent, and it is timely and effectively detect an abnormal electrical behavior in the power system to discover malicious abnormal electrical users, reducing economic losses, ensuring that the power system is safe and reliable, maintaining the steady development of national economies is critical. With the deployment of senior metering infrastructure, massive large-scale high-dimensional complex power consumption data is accumulated, but there are problems of missing value and low abnormal electrical detection accuracy. How to effectively impute the missing value of electricity consumption data to improve data quality, and effectively detect the exceptional electricity user, has been widely concerned in recent years. The purpose of the paper is to improve the accuracy of imputing the power consumption data missing value and the accuracy of detection.

Considering the correlation between high-dimensional power energy data proximity features and potential spatial characteristics, this paper analyzes the electricity consumption data in combination with deep learning, improves abnormal electricity detection accuracy, the main work content is as follows:

(1) Focusing on the problem that the existing missing data imputation model lacks considering neighboring data characteristic correlation, and the missing data imputation value does not meet the real data distribution, the multi-scale sliding window is introduced on the basis of the missing value imputation algorithm. The multi-scale sliding window aims to adaptively extract the characteristics of timing data, capture potential correlation in the neighboring data. We propose a generative adversarial network model with a multi-scale sliding window (MSW-GAN) for missing value imputation. We introduce a prompt matrix in the MSW-GAN generator input to speed up the model training, suppress the discriminator capability. The experiment is carried out on a plurality of data containing timing characteristics, verifying that the proposed model is more accurate to impute the missing value of electricity consumption data.

(2) For the existing electricity theft detection model, the timing characteristics and spatial characteristics of the electricity consumption data set are not taken into account, a mixed deep learning model (CAEs-LSTM) combined of convolutional autoencoder and long short-term memory network is proposed to detect users with abnormal electrical behaviors. The model learns the timing characteristics of electricity consumption data through long short-term memory networks, spatial characteristics through convolutional autoencoder, and perform Gaussian white noise processing to improve model anti-noise capacity. The experiment is conducted by the real data set published by the Power Grids, and the results show that the mixed model effectively improves the detection accuracy.

[1] 张瑶, 王傲寒, 张宏. 中国智能电网发展综述[J]. 电力系统保护与控制, 2021, 49(5): 180-187.

[2] Kosek A M. Contextual anomaly detection for cyber-physical security in smart grids based on an artificial neural network model[C]//2016 Joint Workshop on Cyber-Physical Security and Resilience in Smart Grids (CPSR-SG). IEEE, 2016: 1-6.

[3] Feng Z, Huang J, Tang W H, et al. Data mining for abnormal power consumption pattern detection based on local matrix reconstruction[J]. International Journal of Electrical Power & Energy Systems, 2020, 123: 106315.

[4] 国家发展改革委，国家能源局. 电力发展“十四五”规划[R]. 2021.

[5] 赵文清, 沈哲吉, 李刚. 基于深度学习的用户异常用电模式检测[J]. 电力自动化设备, 2018, 38(9): 34-38.

[6] Henriques H O, Correa R L S, Fortes M Z, et al. Monitoring technical losses to improve non-technical losses estimation and detection in LV distribution systems[J]. Measurement, 2020, 161: 107840.

[7] Glauner P, Meira J A, Valtchev P, et al. The challenge of non-technical loss detection using artificial intelligence: A survey[J]. International Journal of Computational Intelligence Systems, 2017, 10(1): 760-775.

[8] Wu R, Wang L, Hu T. AdaBoost-SVM for electrical theft detection and GRNN for stealing time periods identification[C]//IECON 2018-44th Annual Conference of the IEEE Industrial Electronics Society. IEEE, 2018: 3073-3078.

[9] Hu W, Yang Y, Wang J, et al. Understanding electricity-theft behavior via multi-source data[C]//Proceedings of The Web Conference. 2020: 2264-2274.

[10]Peng Y, Yang Y, Xu Y, et al. Electricity theft detection in AMI based on clustering and local outlier factor[J]. IEEE Access, 2021, 9: 107250-107259.

[11]Jokar P, Arianpoo N, Leung V C M. Electricity theft detection in AMI using customers’ consumption patterns[J]. IEEE Transactions on Smart Grid, 2015, 7(1): 216-226.

[12]Wu Y, Tan X, Qian L, et al. Optimal pricing and energy scheduling for hybrid energy trading market in future smart grid[J]. IEEE Transactions on Industrial Informatics, 2015, 11(6): 1585-1596.

[13]Lin W C, Tsai C F. Missing value imputation: a review and analysis of the literature (2006-2017)[J]. Artificial Intelligence Review, 2020, 53(2): 1487-1509.

[14]谢桃枫, 李宗学. 基于SAS对缺失数据的处理——以新药试验为例[J]. 内蒙古大学学报: 自然科学版, 2017, 48(1): 47-54.

[15]Rumaling M I, Chee F P, Dayou J, et al. Missing Value Imputation for PM 10 Concentration in Sabah using Nearest Neighbour Method (NNM) and Expectation-Maximization (EM) Algorithm[J]. Asian Journal of Atmospheric Environment (AJAE), 2020, 14(1): 62-72.

[16]蒋辉, 马超群, 许旭庆, 等. 仿EM的多变量缺失数据填补算法及其在信用评估中的应用[J]. 中国管理科学, 2019, (3): 11-19.

[17]Karmitsa N, Taheri S, Bagirov A, et al. Missing value imputation via clusterwise linear regression[J]. IEEE Transactions on Knowledge and Data Engineering, 2020, 161: 107840.

[18]郑智泉, 王孟孟, 田维琦. 基于加权K近邻算法的缺失数据填补研究[J]. 智能计算机与应用, 2021, 11(11):31-33+42.

[19]Rashid W, Gupta M K. A perspective of missing value imputation approaches[M]//Advances in Computational Intelligence and Communication Technology. Springer, Singapore, 2021: 307-315.

[20]Razavi-Far R, Cheng B, Saif M, et al. Similarity-learning information-fusion schemes for missing data imputation[J]. Knowledge-Based Systems, 2020, 187: 104805.

[21]Raja P S, Thangavel K. Missing value imputation using unsupervised machine learning techniques[J]. Soft Computing, 2020, 24(6): 4361-4392.

[22]Bertsimas D, Pawlowski C, Zhuo Y D. From predictive methods to missing data imputation: an optimization approach[J]. The Journal of Machine Learning Research, 2017, 18(1): 7133-7171.

[23]Mital U, Dwivedi D, Brown J B, et al. Sequential imputation of missing spatio-temporal precipitation data using random forests[J]. Frontiers in Water, 2020, 2: 20-31.

[24]Zhang X, Yan C, Gao C, et al. Predicting missing values in medical data via XGBoost regression[J]. Journal of Healthcare Informatics Research, 2020, 4(4): 383-394.

[25]Che Z, Purushotham S, Cho K, et al. Recurrent neural networks for multivariate time series with missing values[J]. Scientific reports, 2018, 8(1): 1-12.

[26]Yoon J, Jordon J, Schaar M. Gain: Missing data imputation using generative adversarial nets[C]//International conference on machine learning. PMLR, 2018: 5689-5698.

[27]Luo Y, Cai X, Zhang Y, et al. Multivariate time series imputation with generative adversarial networks[J]. Advances in neural information processing systems, 2018, 31.

[28]Lee D, Kim J, Moon W J, et al. CollaGAN: Collaborative GAN for missing image data imputation[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2019: 2487-2496.

[29]Nalls M A, Blauwendraat C, Vallerga C L, et al. Identification of novel risk loci, causal insights, and heritable risk for Parkinson's disease: a meta-analysis of genome-wide association studies[J]. The Lancet Neurology, 2019, 18(12): 1091-1102.

[30]Guo Z, Wan Y, Ye H. A data imputation method for multivariate time series based on generative adversarial network[J]. Neurocomputing, 2019, 360: 185-197.

[31]Zhang Y, Zhou B, Cai X, et al. Missing value imputation in multivariate time series with end-to-end generative adversarial networks[J]. Information Sciences, 2021, 551: 67-82.

[32]Sardar S, Ahmad S. Detecting and minimizing electricity theft: A review[J]. International Journal on Electrical Innovation., 2015, 4: 121-131.

[33]Yaghmaee M H, Moghaddassian M, Leon-Garcia A. Autonomous two-tier cloud-based demand side management approach with microgrid[J]. IEEE Transactions on Industrial Informatics, 2016, 13(3): 1109-1120.

[34]Biswal M, Dash P K. Measurement and classification of simultaneous power signal patterns with an S-transform variant and fuzzy decision tree[J]. IEEE Transactions on Industrial Informatics, 2012, 9(4): 1819-1827.

[35]Gunturi S K, Sarkar D. Ensemble machine learning models for the detection of energy theft[J]. Electric Power Systems Research, 2021, 192: 106904.

[36]Anwar M, Javaid N, Khalid A, et al. Electricity theft detection using pipeline in machine learning[C]//2020 International Wireless Communications and Mobile Computing (IWCMC). IEEE, 2020: 2138-2142.

[37]Bohani F A, Suliman A, Saripuddin M, et al. A comprehensive analysis of supervised learning techniques for electricity theft detection[J]. Journal of Electrical and Computer Engineering, 2021, 2021.

[38]简富俊, 曹敏, 王磊, 等. 基于SVM 的AMI 环境下用电异常检测研究[J]. 电测与仪表, 2014, 51(6): 64-69.

[39]Kong X, Zhao X, Liu C, et al. Electricity theft detection in low-voltage stations based on similarity measure and DT-KSVM[J]. International Journal of Electrical Power & Energy Systems, 2021, 125: 106544.

[40]Guerrero J I, León C, Monedero I, et al. Improving knowledge-based systems with statistical techniques, text mining, and neural networks for non-technical loss detection[J]. Knowledge-Based Systems, 2014, 71: 376-388.

[41]Yip S C, Tan C K, Tan W N, et al. Energy theft and defective meters detection in AMI using linear regression[C]//2017 IEEE International Conference on Environment and Electrical Engineering and 2017 IEEE Industrial and Commercial Power Systems Europe (EEEIC/I&CPS Europe). IEEE, 2017: 1-6.

[42]Luya L E, Pedrasa M A. Detecting and estimating amount of energy theft in the distribution network using linear regression[C]//2019 9th International Conference on Power and Energy Systems (ICPES). IEEE, 2019: 1-6.

[43]Yip S C, Tan C K, Tan W N, et al. Detection of energy theft and metering defects in advanced metering infrastructure using analytics[C]//2018 International Conference on Smart Grid and Clean Energy Technologies (ICSGCE). IEEE, 2018: 15-22.

[44]Peng Y, Yang Y, Xu Y, et al. Electricity theft detection in AMI based on clustering and local outlier factor[J]. IEEE Access, 2021, 9: 107250-107259.

[45]Hussain S, Mustafa M, James S E, et al. Electric Theft Detection Using Unsupervised Machine Learning-Based Matrix Profile and K-Means Clustering Technique[M]//Computational Intelligence in Machine Learning. Springer, Singapore, 2022: 15-24.

[46]Khanna K, Panigrahi B K, Joshi A. AI-based approach to identify compromised meters in data integrity attacks on smart grid[J]. IET Generation, Transmission & Distribution, 2018, 12(5): 1052-1066.

[47]El Mrabet Z, Selvaraj D F, Nair A S, et al. Detection of the False Data Injection Attack in Home Area Networks using ANN[C]//2019 IEEE International Conference on Electro Information Technology (EIT). IEEE, 2019: 176-181.

[48]Ashrafuzzaman M, Chakhchoukh Y, Jillepalli A A, et al. Detecting stealthy false data injection attacks in power grids using deep learning[C]//2018 14th International Wireless Communications & Mobile Computing Conference (IWCMC). IEEE, 2018: 219-225.

[49]Li S, Han Y, Yao X, et al. Electricity theft detection in power grids with deep learning and random forests[J]. Journal of Electrical and Computer Engineering, 2019: 1-12.

[50]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, 2017, 14(4): 1606-1615.

[51]Srindhuna M, Baburaj M. Estimation of missing data in remote sensing images using t-SVD based tensor completion[C]//2020 International Conference for Emerging Technology (INCET). IEEE, 2020: 1-5.

[52]Kakkar M, Jain S, Bansal A, et al. Evaluating Missing Values for Software Defect Prediction[C]//2019 International Conference on Machine Learning, Big Data, Cloud and Parallel Computing (COMITCon). IEEE, 2019: 30-34.

[53]Santos M S, Pereira R C, Costa A F, et al. Generating synthetic missing data: A review by missing mechanism[J]. IEEE Access, 2019, 7: 11651-11667.

[54]Li X, Li G, Fishbune R. A novel missing-rate-oriented selective algorithm for handling missing data by minimizing imputation[C]//2016 International Conference on Cyber-Enabled Distributed Computing and Knowledge Discovery (CyberC). IEEE, 2016: 234-237.

[55]Sobrevilla K L M D, Quiñones A G, Lopez K V S, et al. Daily weather forecast in Tiwi, Albay, Philippines using artificial neural network with missing values imputation[C]//2016 IEEE Region 10 Conference (TENCON). IEEE, 2016: 2981-2985.

[56]Finardi P, Campiotti I, Plensack G, et al. Electricity theft detection with self-attention[J]. Journal of Electrical and Computer Engineering, 2020: 1-12.

[57]Rehman F, Abbas M, Murtaza S, et al. SimFiller. Similarity-Based Missing Values Filling Algorithm[C]//2018 Thirteenth International Conference on Digital Information Management (ICDIM). IEEE, 2018: 77-81.

[58]Tripathi A K, Rathee G, Saini H. Taxonomy of missing data along with their handling methods[C]//2019 Fifth International Conference on Image Information Processing (ICIIP). IEEE, 2019: 463-468.

[59]Kundu A, Sahu A, Davis K, et al. Learning-based defense of false data injection attacks in power system state estimation[C]//2019 North American Power Symposium (NAPS). IEEE, 2019: 1-6.

[60]Niu X, Li J, Sun J, et al. Dynamic detection of false data injection attack in smart grid using deep learning[C]//2019 IEEE Power & Energy Society Innovative Smart Grid Technologies Conference (ISGT). IEEE, 2019: 1-6.

[61]Wang L, Zhou Q. Physics-guided deep learning for time-series state estimation against false data injection attacks[C]//2019 North American Power Symposium (NAPS). IEEE, 2019: 1-6.

[62]Hu T, Guo Q, Sun H, et al. Nontechnical losses detection through coordinated BiWGAN and SVDD[J]. IEEE Transactions on Neural Networks and Learning Systems, 2020, 32(5): 1866-1880.

[63]I. Goodfellow, Y. Bengio, A. Courville. Deep Learning. Massachusetts, USA :MIT Press, 2016: 23-36.

[64]Chang Z, Zhang Y, Chen W. Electricity price prediction based on hybrid model of adam optimized LSTM neural network and wavelet transform[J]. Energy, 2019, 187: 115804.

[65]Le T, Vo M T, Vo B, et al. Improving electric energy consumption prediction using CNN and Bi-LSTM[J]. Applied Sciences, 2019, 9(20): 4237.

[66]COLAK I, SAGIROGLU S, FULLI G, et al. A survey on the critical issues in smart grid technologies[J]. Renewable and Sustainable Energy Reviews, 2016, 54: 396-405.

[67]Vincent P, Larochelle H, Bengio Y, et al. Extracting and composing robust features with denoising autoencoders[C]//Proceedings of the 25th international conference on Machine learning. 2008: 1096-1103.

[68]Cheng H T, Koc L, Harmsen J, et al. Wide & deep learning for recommender systems[C]//Proceedings of the 1st workshop on deep learning for recommender systems. 2016: 7-10.

[69]Huang Y, Xu Q. Electricity theft detection based on stacked sparse denoising autoencoder[J]. International Journal of Electrical Power & Energy Systems, 2021, 125: 106448.