基于智能电表大数据的异常用电检测 |
投稿时间:2019-05-26 修订日期:2019-06-25 点此下载全文 |
引用本文: |
摘要点击次数: 76 |
全文下载次数: 0 |
|
|
中文摘要:异常或欺诈造成的非技术性电力损失是电力公司损失的主要源头之一。智能电表的广泛使用,使得运用大数据方法实现对非技术性电力损失的有效检测成为可能。为此本文提出了一种使用监督学习进行非技术损失检测的方法。该方法基于智能仪表记录的所有信息(耗电量、异常警报等)结合辅助数据库所提供的有关每个智能电表的地理位置和技术参数的附加信息,使用最优的机器算法来深入分析用电客户的用电行为,生成异常用电客户列表。通过现场检查的结果表明,该方法能够较为准确地识别智能电网中所存在异常用电客户。 |
中文关键词:监督学习,非技术损失,智能电表,超梯度提升树 |
|
Abnormal power detection based on smart meter big data |
|
|
Abstract:Non-technical power loss is one of the main sources of power company abnormal losses. The wide use of intelligent meters makes it possible to effectively detect non-technical power losses by using big data. In this paper, a supervised learning method for non-technical loss detection is proposed. Based on all the information recorded by intelligent meters (power consumption, abnormal alarm, etc.) and the additional information about the geographical location and technical parameters of each smart meter provided by the auxiliary database, the method uses the optimal machine algorithm to analyze the power consumption behavior of power customers in depth, and generates a list of abnormal power users. The results of on-site inspection show that this method can accurately identify abnormal customers in smart grid. |
keywords:Supervised Learning, Non-Technical Loss, Smart Meter, Extreme Gradient Boosted Trees |
查看全文 查看/发表评论 下载pdf阅读器 |