基于电力大数据及孤立森林算法的窃电识别研究  

Research on Electricity Theft Identification Based on Electric Power Big Data and Isolation Forest Algorithm

作　　者：张明艳 陈昕 晋成凤 ZHANG Mingyan;CHEN Xin;JIN Chengfeng(State Grid Ma’anshan Power Supply Company,Ma’anshan 243000,China;Guodian Nanjing Automation Co.,Ltd,Nanjing 211100,China)

机构地区：[1]国网马鞍山供电公司,安徽马鞍山243000 [2]国电南京自动化股份有限公司,江苏南京21110

出　　处：《山东电力高等专科学校学报》2025年第1期35-38,共4页Journal of Shandong Electric Power College

摘　　要：针对家庭用户窃电行为,在分析现有反窃电技术的基础上,提出基于电力大数据及孤立森林算法的窃电识别方法。该方法结合工作中遇到的家庭用户窃电案例,确定窃电行为判别指标,通过熵权法计算判别指标的权值,并应用孤立森林算法建立窃电检测模型。最后进行模型仿真实验,并将实验结果与其他方法构建的模型进行比较,结果表明该方法对窃电行为识别的准确度较高、漏警率较低,具有较好的实用价值。An electricity theft identification method based on the electric power big data and isolation forest algorithm is proposed,after analyzing the existing electricity theft prevention technology,so as to cope with the electricity theft behavior of household users.The method determines the indexes for identifying electricity theft behavior based on electricity theft cases of household users encountered in work,calculates the weights of the identification indexes through the entropy weight method,and uses the isolation forest algorithm to establish an electricity theft detection model.Finally,the model simulation experiment is conducted,and the experimental results are compared with the models constructed by other methods,which proves that the method has high accuracy and low missed alarm rate in identifying electricity theft behavior,and has good practical value.

关 键 词：电力大数据 窃电行为判别指标 熵权法 孤立森林算法 

分 类 号：TM73[电气工程—电力系统及自动化]