The accuracy of user-transformer relationship identification plays a key role in the safe and stable operation of the distribution network, and its results directly affect the accuracy of line loss calculation of the distribution network. The traditional distribution station recognition algorithm cluster the power frequency zero-crossing sequence, which does not perform well when the sequence is ‘non-convex. Accordingly, an improved spectral clustering algorithm for arbitrary zero-crossing sequences is proposed. The algorithm takes maximizing the variance of the weight matrix (WM) as the objective function of the adaptive particle swarm optimization algorithm, then adaptively selects the parameter threshold of WM to change the WM into a sparse matrix. Thus, the eigenvectors calculation problem of the traditional spectral clustering algorithm is simplified to find the orthogonal null space of the canonical Laplace matrix. The recognition accuracy of the proposed algorithm with sample data obtained by the simulation software can reach 99.11%, which is better than that of the traditional clustering algorithm.

Adaptive particles swarm optimization, Maximize variance, null space, Spectral clustering, User-transformer relationship identification