摘要
常规的终端用户用电行为异常检测,主要采用数据偏置量幅度进行异常检测,忽略了空值数据对检测结果的影响。因此,提出基于电力大数据的终端用户用电行为异常检测研究。剔除空值占比在50%以上的用户用电行为数据,采用牛顿插值法和离差标准化对剔除后的数据进行处理,结合电力大数据对用电行为异常数据特征进行分析,判定异常并分类得出异常检测结果。实验结果表明:所提方法应用后得出的检测结果,表现出的ROC(Receiver Operating Characteristic)曲线较凸,得到的AUC量化值更加趋近于1,准确度较高,可满足电网终端运维的现实需求。
The conventional abnormal detection of end-user’s power consumption behavior mainly adopts the data bias amplitude to detect the anomaly,ignoring the influence of null data on the detection result.Therefore,this paper proposes a study on the abnormal detection of end-user’s power consumption behavior based on power big data.The consumption behavior data of users with null value ratio of more than 50%are eliminated,Newton interpolation method and deviation standardization are used to process the eliminated data,and the characteristics of abnormal consumption behavior data are analyzed in combination with power big data,and abnormal detection results are determined and classified.The experiment results show that the detection results obtained after the application of the proposed method show a more convex curve of the Receiver Operating Characteristic(ROC),and the quantized value of the AUC obtained is closer to 1,with high accuracy,which meets the practical needs of power grid terminal operation and maintenance.
作者
洪莹
王圣竹
王缉芬
HONG Ying;WANG Sheng-zhu;WANG Ji-fen(Customer Service Center of Guangxi Power Grid Co.,Ltd.,Nanning 530000,China)
出处
《信息技术》
2025年第2期144-149,155,共7页
Information Technology
关键词
终端用户
用电行为异常
检测方法
空值占比
牛顿插值法
end user
abnormal use of electricity
detection method
null ratio
Newton interpolation
