Identification of User-transformer Relationship Based on Dynamic Time Warping Algorithm and Two-stage Clustering Algorithm

With the continuous promotion of new power system construction, a higher requirement is proposed for low-voltage district management. Problems such as abnormal data collection, line loss caused by user-transformer relationship chaos aren’t to be ignored. To solve the problem, this paper proposed a two-stage clustering algorithm for user-transformer relationship identification. First, the similarity of users’ voltage time series curve is calculated by dynamic time warping algorithm. Second, propose a two-stage clustering algorithm based on the self-organizing feature mapping algorithm and K-means algorithm. With this two- stage clustering algorithm, the similarity of users’ voltage time series curve could be clustered rapidly. Users connected to the same transformer that have similar voltage time series curve theoretically could be identified to be the same type, therefore achieving the goal of identifying user-transformer relationship intelligently. Finally, simulations and analysis are performed on the basis of actual low-voltage district. Results prove the effectiveness of the method proposed in this paper.