Data Analysis and Data Generation Techniques for Comparative Examination of Distribution Network Topologies

Abstract

Nowadays photovoltaic systems can be considered to play the key role in increasing decentralized power generation. These power generation systems typically connect to low and medium voltage distribution networks, thus affecting their operation. The proliferation of their increasing number and high power connection demand the transformation of the operation of the distribution network and the development of the infrastructure. An examination of development opportunities is in most cases done by software simulation, which requires network models to be created. In order to be able to formulate rules of thumb for such developments, a large-scale simulation is needed. To achieve this, the generation of reference networks that describe the real networks well is essential. In order for the reference networks to be created, clustering of real networks is required. When the clusters and their properties are properly defined and available, then the software implementable networks can be created. The aim of this paper is to give a brief overview of the clustering methodologies of radial distribution networks (including statistical clustering, dimensional reduction, agglomerative clustering, partition clustering, K-means and K-medoids clustering) and examine which may be useful for the comparison of advantages and disadvantages of these processes.
Copyright © 2019 Praise Worthy Prize - All rights reserved.