Connection pattern recognition for medium-voltage distribution network in power supply area

Inhalt:
Connection pattern recognition is the basis of distribution system planning and evaluation. Integrated with a connection pattern model for 10kV urban medium-voltage distribution networks, here proposes a connection pattern recognition method based on topology analysis and clustering algorithm. For analysis the network structure, the depth-first search algorithm is adopted to divide a complete network into outlet trees, which are the research object of connection pattern recognition. Typical power grid structures of 10kV distribution network are analyzed, including overhead network and cable network. On this basis, the features of typical connection modes belonging to each grid structure are extracted and analyzed respectively, and feature expressions as well as feature database are established to cover all typical modes. Finally, the clustering algorithm is utilized to match the features of unknown connection modes with the centers of class, whose cluster centers are composed with the typical modes. The recognition results of an example network validate the feasibility of the proposed algorithm, which has guiding significance for the planning and construction of medium-voltage distribution networks.