Formulation of Reference Network Models for the Hungarian Medium Voltage Distribution Level

In the next two decades, a rapidly increasing photovoltaic penetration connected to the distribution electricity network is expected both nationally and internationally, which influences grid operation. Simultaneously, an essential development of network infrastructure is also expected. In order to determine development directions and opportunities, power system simulation is essential, which is based on the generation of representative feeders. In this paper, by clustering (using k-means clustering) medium voltage distribution feeders, valid Hungarian representative networks are formulated and described. The number of descriptive variables was reduced by, principal component analysis. The feeders are described with the first three principal components (95% coverage of the information). The optimal number of clusters was determined using four different validity criteria, and set to 6. The contributions of the paper are that the examined feeders are characterized by both electrical (physical) and mathematical parameters, in a new combination, from which the number of switches is a novel element in the literature. The applied principal component analysis works with more components (3) and greater coverage (95%) than the methods available in the examined literature. Based on topographic verification of the representative networks, the clusters are well characterized by the parameters of the feeders.