Graph partitioning-based clustering for the planning of distribution network topology using spatial- temporal load forecasting

Nenhuma Miniatura disponível




Zambrano-Asanza, S. [UNESP]
Cando, Diego J.
Chuqui, Freddy H.
Sanango, Juan
Franco, John F. [UNESP]

Título da Revista

ISSN da Revista

Título de Volume



Planning the expansion and the new topology of distribution networks requires knowing the location and characterization of the load as well as its future growth. Spatial load forecasting is a key tool in this task, providing high spatial resolution and adequate temporal granularity. Nowadays, with the penetration of distributed energy resources, multiple microgrid connection strategies, and implementation of self-healing and protection schemes, it is necessary to identify load blocks to plan the new active network architecture. Based on spatial load forecasting information, this paper proposes a graph partitioning technique to create load clusters in the distribution feeders. A weighted graph is constructed by means of a minimum spanning tree that allows to consider adjacency relations. The results of the simulation, carried out in a real distribution network, have demonstrated the effectiveness of the proposed method.



Clustering, Distribution planning, Graph partitioning, Microgrids, Minimal spanning tree, Spatial load forecasting

Como citar

2021 IEEE PES Innovative Smart Grid Technologies Conference - Latin America, ISGT Latin America 2021.