基于改进K-Means++聚类分析的邻户表计错接辨识方法

Neighbor Meter Misconnection Identification Method Based on Improved K-Means++Clustering Analysis

邻户间表计错接影响电能计量的准确性,致使电费结算异常。人工排查方式效率低、难度大,基于用户电量数据,通过改进K均值聚类分析提出了一种单相表计邻户错接线辨识方法。首先,从SG186系统与采集系统中提取用户换表信息与日用电量数据;其次,通过均值、标准差、峭度3个统计量刻画用户换表前后的日用电量数据段曲线特征,并在多维空间内标定出每段数据的特征点;最后,对客户换表前后用电数据段的特征点进行K-means++聚类分析以错接辨识邻户表计错接情况。实际案例应用分析验证了所提方法的有效性,辨识结果可作为错接判别的有效依据。

The misconnection of meters between neighbors affects the accuracy of power metering,resulting in abnormal settlement of electricity charges.The manual checking method is inefficient and difficult.Based on the consumers'electricity data,a method for identifying the wrong connection of neighbors with single-phase meter is proposed through the improved K-means cluster analysis.Firstly,the data of customers'meter change and daily electricity consumption are extracted from SGi86 system and collection system.Secondly,the curve characteristics of the daily electricity consumption data segment before and after the customer meter change are described by means of the mean,standard deviation and kurtosis statistics,and the feature points of each data segment are calibrated in the multidimensional space.Finally,K-Means++cluster analysis is carried out on the characteristic points of the power consumption data segment before and after the customer meter changes to identify the misconnection.The effectiveness of the proposed method is verified by practical case analysis,and the identification results can be used as an effective basis for misconnection discrimination.