基于t-SNE降维和BIRCH聚类的单相用户相位及表箱辨识

Phase and Meter Box Identification for Single-phase Users Based on t-SNE Dimension Reduction and BIRCH Clustering

低压台区单相用户的相位及接入表箱信息的准确性对户变关系纠错和线损治理分析有重要影响。目前,拓扑档案的校验主要依靠电力员工现场排查,人力物力消耗大且排查效率低下。因此,亟需一种效率较高的低压台区拓扑档案校验方法。在此背景下,文中提出了一种基于智能电表电压数据的低压台区单相用户相位及接入表箱辨识方法,可以为低压台区的拓扑辨识及排查提供参考。首先,采用t分布的随机近邻嵌入(t-SNE)技术对原始负荷数据进行降维处理,解决台区用户原始负荷特征维度过高带来的冗余性问题;接着,应用BIRCH方法对降维后的负荷数据进行聚类,实现台区下单相用户所属相位和接入表箱的辨识。最后,以浙江省海宁市某台区为例进行验证,算例分析的结果表明所提模型具有可行性和有效性。

The accurate phase and meter box information of single-phase users in low-voltage courts have great impacts on the checking of user-transformer relationships and the treatment and analysis of line losses.At present,the correction of topological documents mainly relies on the on-site checking by electrical engineers,which spends a lot of manpower and material resources with low efficiency.Therefore,it is necessary to study a more efficient method of checking the topological documents of the lowvoltage courts.In this background,a phase and meter box identification method based on voltage measurement data of smart meters is proposed,which provides reference for topology identification and correction of low-voltage courts.Firstly,the tdistributed stochastic neighbor embedding(t-SNE)technology is adopted to reduce the dimension of original load data,so as to solve the redundancy problems caused by excessively high dimension of original load characteristics of users.Then,the balanced iterative reducing and clustering using hierarchies(BIRCH)method is used to cluster the dimension-reduced load data,so as to identify the phase and meter box information of single-phase users.Finally,case studies for an actual low-voltage court in Haining of Zhejiang Province,China,are performed to verify the correctness of the proposed method,and the results show that the proposed model is feasible and effective.