基于非负矩阵分解和改进相关分析的低压台区拓扑辨识方法

Topology Identification Method for Low-voltage Substation Area Based on Non-negative Matrix Factorization and Improved Correlation Analysis

针对低压配电台区拓扑更新不及时,量测质量低、拓扑信息不够精确的问题,提出一种基于非负矩阵分解改进Pearson相关系数的低压配电台区拓扑辨识方法。首先,利用非负矩阵分解对电压时间序列进行降维。然后,在Pearson相关系数的基础上,加入距离度量来修正相似性矩阵。最后,对综合相似度矩阵进行系统聚类,实现低压配电台区户-变识别和台区内用户相邻关系识别。通过实际低压配电台区算例验证了所提方法的有效性和准确性。

Aimed at the problems such as untimely update of low-voltage distribution substation area topology files,low quality of metering and inaccurate topological information,a topology identification method for low-voltage distribution substation area based on non-negative matrix factorization(NMF)and improved Pearson correlation coefficient is pro⁃posed in this paper.First,NMF is used to reduce the number of dimensions of the voltage time series data.Second,on the basis of Pearson correlation coefficient,a distance metric is added to correct the similarity matrix.Finally,system clustering is performed on the comprehensive similarity matrix,thus realizing the identification of user-transformer in low-voltage distribution substation areas and the refined identification of user adjacencies within these areas.The effec⁃tiveness and accuracy of the proposed method are verified by an actual low-voltage distribution substation area as a cal⁃culation example.