摘要
目前有效的用电行为异常监测手段,仅依靠计量人员、用电检查人员逐户现场核查,导致工作效率低、工作量大,用户下场检查覆盖率无法达到100%。为解决该问题,本文以逆向思维为主线,从历史计量差错、电价执行异常、违约、窃电用户入手,提炼阈值规则,搭建客户用电异常行为精准识别模型,自动校验输出表计失压、断流、接线异常、私自启封、私自增容、高低压电价混接等监测场景异动;同时基于行业信息、用电时段、电压、电流、电量等数据,通过应用大数据技术进行回归分析、K-means聚类分析等计算,构建用电客户行业特征模型及电力波动模型,并与电量突增突减、台区线损、表计开盖规则场景异动精准率,辅助计量、用检人员提高下厂检查效率和电费审核效率,助力电费“颗粒归仓”。
At present, effective abnormal monitoring means of electricity consumption behavior only rely on metering personnel and electricity consumption inspection personnel for on-site verification, resulting in low efficiency and heavy workload, and the coverage rate of users’ follow-up inspection cannot reach 100%.In order to solve this problems, the research direction of this paper is the reverse thinking as the main line, starting with historical metering errors, abnormal electricity price execution, default, electricity stealing users, refining threshold rules, automatic verification of output meter voltage loss, cut off, wiring abnormal, unsealed, unexpanded capacity, high and low voltage electricity price mixed monitoring scenes;at the same time, based on industry information, power consumption period, voltage, current, electricity and other data, through the application of big data technology for regression analysis, K-means clustering analysis and other calculations, the construction of customer industry characteristic model and power fluctuation model, and with the sudden increase and decrease of electricity, station line loss, meter cover opening rules of the scene change accuracy rate.Assist metering and inspectors to improve inspection efficiency and electricity audit efficiency, and help electricity “grain warehousing”.
作者
陈明
张丽文
王璐
袁娟
宋庆华
曾琴
CHEN Ming;ZHANG Wenll;WANG Lu;YUAN Juan;SONG Qinghua;ZENG Qin(State Grid Jiuquan Power Supply Company of Gansu Electric Power Company,Jiuquan 735000,Gansu,China;Chengdu Kepuwei Information Technology Co.,Ltd.,Chengdu 610042,Sichuan,China)
出处
《电力大数据》
2022年第6期24-35,共12页
Power Systems and Big Data
关键词
数据挖掘
全量数据
异常用电行为
行业特征
场景监测
data mining
full data
abnormal power consumption behavior
industry characteristics
scene monitoring
