Method for Consumer-transformer Relationship Identification Based on Diverse data

Abstract

This paper proposes a method for identifying consumer-transformer relationships based on voltage fluctuation feature clustering and power summation relationships. First, the voltage fluctuation characteristics of the distribution station area are analyzed, the set of feature parameters reflecting the global and local characteristics of voltage fluctuation is defined, and its calculation method is given. Then, the multi-dimensional fluctuation parameters of the neighboring substation voltages and their customer voltages are extracted and the t-distributed stochastic neighbor embedding (t-SNE) algorithm is used to reduce the dimensionality of the high-dimensional voltage fluctuation features. Finally, based on the summation relationship between the virtual user power data and the concentrator, the least-squares method is used for pairing to identify the household-variable relationship and the phase of users in the distribution network. Compared with the traditional topology identification method which directly uses voltage data for clustering, the proposed method has a higher identification accuracy and solves the problem that the traditional method cannot correctly identify when the voltages of different station areas are close to each other.