基于聚类相关性分析的台区用电异常用户查找方法

A Method for Finding Abnormal Electricity Users in Substation Areas Based on Cluster Correlation Analysis

现有的查窃技术手段有特定的应用场景,线损异常台区因现有查窃技术手段无法锁定用电异常用户,无法进一步降损。实践发现,单用户的用电行为与台区线损不存在直接相关性,而用电异常用户整体的用电行为却与台区线损高度相关,因此,本文利用计量自动化系统中的用电数据,采用聚类相关性分析算法找出其负荷曲线与台区损失电量曲线相关性最高的用户集合,用于锁定用电异常用户群。实验证明,该方法能准确锁定用电异常用户群,与台区全覆盖式用电检查相比,查窃效率提高数倍,且能有效锁定用电行为、用电负荷曲线特征无明显异常的用电异常用户。

The existing theft detection technology has specific application scenarios,and the abnormal line loss substation cannot further reduce losses due to the inability of the existing theft detection technology to lock in users with abnormal electricity consumption.Practice has found that there is no direct correlation between the electricity consumption behavior of a single user and the line loss in the substation area,while the overall electricity consumption behavior of users with abnormal electricity consumption is highly correlated with the line loss in the substation area.Therefore,this paper uses electricity consumption data from the metering automation system and uses clustering correlation analysis algorithm to find the user set with the highest correlation between their load curve and the line loss curve in the substation area,used to lock in user groups with abnormal electricity usage.Experiments have shown that this method can accurately identify abnormal electricity users.Compared to the full coverage electricity inspection in the substation area,the efficiency of theft detection is several times higher,and it can effectively identify abnormal electricity users whose electricity consumption behavior and electricity load curve characteristics are not significantly abnormal.