高维流形视角下采用ISOMAP降维的配网户变关系辨识

准确的户变关系是配网线损计算、故障定位和三相平衡等高级应用的基础.低压配网的户变关系辨识算法大多基于电压相关性原理,而电压相关性随供电半径增加而减弱,电压采集频次较低无法可靠捕获电压的"共性波动",使得辨识准确率普遍不高.提出了一种基于等距特征映射(isometric mapping,ISOMAP)降维和改进K-mea...

Identification of user-transformer relationship using ISOMAP method from high dimensional flow pattern perspective

Accurate user-transformer relationship is regarded as the basis of advanced applications such as line loss calculation, fault location and three-phase imbalance analysis. The identification algorithm of user-transformer relationship in power distribution is mainly based on the principle of voltage correlation. Due to the voltage correlation weakens with the increase of power supply radius, the "common fluctuation" of voltage can not be captured reliably with lower acquisition frequency. The low identification accuracy is caused. In this paper, an identification algorithm of user-transformer relationship based on ISOMAP and improved K-means is proposed. Firstly, the local communication module of smart meter based on NB-IoT is designed to extend the time scale of voltage sequence. By optimizing the acquisition architecture, the voltage sequence is increased to 288 points each day. Secondly, the topological relationship between nodes is regarded as a high-dimensional flow pattern, and ISOMAP algorithm is used to reduce the dimension of the voltage matrix. Finally, the K-means algorithm is improved by using the geodesic distance, and the identification result of user-transformer relationship is obtained by clustering. The confidence of the distance between nodes is improved by proposed method. The results show that the identification accuracy of user-transformer relationship can reach 97.1% compared with PCA and K-means algorithm, respectively. The identification effectiveness of the proposed algorithm is verified by the actual operation data set of distribution network.