Abstract:Due to the improvement of power supply reliability and the increase of power users, the scale and complexity of distribution network are increasing, which increases the difficulty and time of distribution network topology. Therefore, in order to speed up the topology identification, an improved matrix algorithm is proposed, which divides the topology into global topology and local topology, and improves the two topologies respectively. For global topology, node elimination method is used for adjacency matrix, row scanning method is used for connected matrix, and twice node elimination method is used for incidence matrix to form local topology correlation matrix. After global topology, local topology is carried out for displacement switch. Firstly, according to the change of remote signal, it is known which switches in the network are shifted, and then according to local topology correlation matrix, the connection relationship between displacement switches and local network nodes is obtained. Finally, according to these information, the connection matrix is modified locally, and then the line scanning, bus analysis and electrical island analysis of the matrix are carried out to obtain the local topology of the network, which reflects the global topology of the network. The proposed method has a high reliability and wide application range, and can fully improve the efficiency of topology identification.