基于堆叠去相关自编码器和支持向量机的窃电检测

被引：85

作者：

胡天宇 ^{[1
]}

郭庆来 ^{[1
,2
]}

孙宏斌 ^{[1
,2
]}

机构：

[1] 清华-伯克利深圳学院

[2] 清华大学电机工程与应用电子技术系

来源：

电力系统自动化 | 2019年 / 43卷 / 01期

基金：

国家重点研发计划;

关键词：

非技术性损失; 窃电检测; 深度学习; 去相关自编码器; 支持向量机;

D O I：

暂无

中图分类号：

TM73 [电力系统的调度、管理、通信];

学科分类号：

摘要：

已有窃电检测模型的准确率尚无法满足应用需求,是因其均将建模重点放在了分类算法的选择或改进上,而相对地忽视了特征提取过程。因此,提出一种基于深度学习的特征提取方法,即堆叠去相关自编码器。得益于深层结构和高度非线性,其能够从用户用电数据中提取到高度抽象和简明的特征。随后支持向量机将这些特征映射到指示是否窃电的标签。基于真实数据的算例测试,验证了所提窃电检测模型具有较高的检出率和较低的虚警率,同时也验证了堆叠去相关自编码器能够提取到有效的特征。

引用

页码：119 / 125

页数：7

共 15 条

[1] 基于深度学习和蒙特卡洛树搜索的机组恢复在线决策
孙润稼
刘玉田
[J]. 电力系统自动化, 2018, 42 (14) : 40 - 47
[2] 风险调度中引入知识迁移的细菌觅食强化学习优化算法
韩传家
张孝顺
余涛
瞿凯平
[J]. 电力系统自动化, 2017, 41 (08) : 69 - 77+97
[3] Energy-Theft Detection Issues for Advanced Metering Infrastructure in Smart Grid[J]. Rong Jiang,Rongxing Lu,Ye Wang,Jun Luo,Changxiang Shen,Xuemin(Sherman) Shen.Tsinghua Science and Technology. 2014(02)
[4] On the Study of Commercial Losses in Brazil: A Binary Black Hole Algorithm for Theft Characterization[J] . Ramos Caio C. O.,Rodrigues Douglas,de Souza Andre N.,Papa Joao P..IEEE Transactions on Smart Grid . 2018 (2)
[5] Electricity Theft Detection in AMI Using Customers' Consumption Patterns
Jokar, Paria
Arianpoo, Nasim
Leung, Victor C. M.
[J]. IEEE TRANSACTIONS ON SMART GRID, 2016, 7 (01) : 216 - 226
[6] Decision Tree and SVM-Based Data Analytics for Theft Detection in Smart Grid
Jindal, Anish
Dua, Amit
Kaur, Kuljeet
Singh, Mukesh
Kumar, Neeraj
Mishra, S.
[J]. IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS, 2016, 12 (03) : 1005 - 1016
[7] New Insights on Nontechnical Losses Characterization Through Evolutionary-Based Feature Selection
Oba Ramos, Caio Cesar
de Souza, Andre Nunes
Falcao, Alexandre Xavier
Papa, Joao Paulo
[J]. IEEE TRANSACTIONS ON POWER DELIVERY, 2012, 27 (01) : 140 - 146
[8] A novel algorithm for feature selection using Harmony Search and its application for non-technical losses detection
Ramos, Caio C. O.
Souza, Andre N.
Chiachia, Giovani
Falcao, Alexandre X.
Papa, Joao P.
[J]. COMPUTERS & ELECTRICAL ENGINEERING, 2011, 37 (06) : 886 - 894
[9] Detection and Identification of Abnormalities in Customer Consumptions in Power Distribution Systems
dos Angelos, Eduardo Werley S.
Saavedra, Osvaldo R.
Carmona Cortes, Omar A.
de Souza, Andre Nunes
[J]. IEEE TRANSACTIONS ON POWER DELIVERY, 2011, 26 (04) : 2436 - 2442
[10] Improving SVM-Based Nontechnical Loss Detection in Power Utility Using the Fuzzy Inference System
Nagi, Jawad
Yap, Keem Siah
Tiong, Sieh Kiong
Ahmed, Syed Khaleel
Nagi, Farrukh
[J]. IEEE TRANSACTIONS ON POWER DELIVERY, 2011, 26 (02) : 1284 - 1285

← 1 2 →