基于无监督学习的电力用户异常用电模式检测.pdfVIP

第 36 卷 第 2 期 中 国 电 机 工 程 学 报 Vol.36 No.2 Jan.20, 2016 2016 年 1 月 20 日 Proceedings of the CSEE ©2016 Chin.Soc.for Elec.Eng. 379 DOI ：10.13334/j.0258-8013.pcsee.2016.02.008 文章编号：0258-8013 (2016) 02-0379-09 中图分类号：TM 715 基于无监督学习的电力用户异常用电模式检测 1 2 1 3 1 庄池杰 ，张斌 ，胡军 ，李秋硕 ，曾嵘 (1．电力系统及发电设备控制和仿真国家重点实验室(清华大学电机系) ，北京市 海淀区 100084 ； 2．国家电网公司西北分部，陕西省 西安市 7100481; 3．南方电网科学研究院，广东省 广州市 510080) Anomaly Detection for Power Consumption Patterns Based on Unsupervised Learning 1 2 1 3 1 ZHUANG Chijie , ZHANG Bin , HU Jun , LI Qiushuo , ZENG Rong (1. State Key Lab of Control and Simulation of Power Systems and Generation Equipments (Dept. of Electrical Engineering, Tsinghua University), Haidian District, Beijing 100084, China; 2. Northwest Branch of State Grid Corporation China, Xi’an 710048, Shaanxi Province, China; 3. CSG Electric Power Research Institute, Guangzhou 510080, Guangdong Province, China) ABSTRACT: The primary purpose of anomaly detection for 度排序靠前的少数用户即可查出大部分异常用户。 power consumption patterns is to lower the non-technical 关键词：用电模式；电力大数据；异常检测；无监督学习； losses (NTL), thus reducing the operating costs for power 局部离群因子;反窃电技术 utility. A model based on unsupervised learning was proposed to detect anomaly consumption patterns. This model is 0 引言 suitable for load dataset without training set. The model 随着电力系统信息化程度的不断提高和配用 includes modules of feature extraction, principal component