摘要
传统用电行为特征提取方法难以挖掘连续的用户用电行为数据,导致特征提取精度较低且特征异常值较多。针对这个问题,文章研究电价审计中用户用电行为特征的智能提取方法。挖掘周期性电价审计过程中存在的隐藏数据,标定数据参数,归一化处理用户用电行为数据。采用线性差值填充算法处理周期范围内的电力数据,控制用户用电数据的连续性。特征化处理行为数据后,进行卡方检验并构建指标提取过程,最终完成智能提取方法的构建。在模拟用户用电行为数据采集环境后,对应不同的采集周期整理为用电数据集,准备2种传统提取方法与所设计的方法进行对照实验,实验结果表明,所设计的智能提取方法具有较高的特征提取精度,且产生的特征异常值最少。
The traditional feature extraction method of power consumption behavior is difficult to mine continuous user power consumption behavior, resulting in low feature extraction accuracy and many feature outliers. Aiming at this problem, anintelligent feature extraction method of user power consumption behavior in electricity price audit is studied. The method mines the hidden data existing in the periodic electricity price audit process, calibrates the data parameters, and normalizes the user’s electricity consumption behavior data. The linear difference filling algorithm is used to process the power data within the cycle range to control the continuity of user power data. After characterizing the behavior data, the χ~2 test is carried out and the index extraction process is constructed.Finally, the construction of intelligent extraction method is completed. After simulating the user’s electricity behavior data acquisition environment, it is sorted into electricity consumption data sets corresponding to different acquisition cycles, and two traditional extraction methods and the designed extraction methods are prepared for experiments. The experimental results show that the designed intelligent extraction method has high feature extraction accuracy and produces the least feature outliers.
作者
王兰君
王威
WANG Lanjun;WANG Wei(State Grid Ocean Power Company,Shanghai 200135,China)
出处
《微型电脑应用》
2023年第2期157-160,共4页
Microcomputer Applications
关键词
电价审计
用户用电行为特征
智能提取方法
数据连续性
electricity tariff audit
characteristics of customers’electricity consumption behavior
intelligent extraction method
data continuity
