Energy profiling of end-users in service and industry sectors with use of Complex Network Analysis

Decarbonization scenarios advocate the transformation of energy systems to a decentralized grid of prosumers. However, in heterogeneous energy systems, profiling of end-users is still to be investigated. As a matter of fact, the knowledge of electrical load dynamics is instrumental to the system eff...

Descripción completa

Guardado en:
Detalles Bibliográficos
Autores principales: Portera Rosario, Bonacina Fabrizio, Corsini Alessandro, Miele Eric Stefan, Ricciardi Celsi Lorenzo
Formato: article
Lenguaje:EN
FR
Publicado: EDP Sciences 2021
Materias:
Acceso en línea:https://doaj.org/article/ade162896eab4f39b7c4f2ca047e4ea3
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
Descripción
Sumario:Decarbonization scenarios advocate the transformation of energy systems to a decentralized grid of prosumers. However, in heterogeneous energy systems, profiling of end-users is still to be investigated. As a matter of fact, the knowledge of electrical load dynamics is instrumental to the system efficiency and the optimization of energy dispatch strategies. Recently, a number of clustering algorithms have been proposed to group load diagrams with similar shapes, generating typical profiles. To this end, conventional clustering algorithms are unable to capture the temporal dynamics and sequential relationships among data. This circumstance is of paramount importance in the service and industrial sectors where energy consumption trends over time are possibly non-stationary. In this paper, we aim to reconstruct the annual user energy profile identified through a non-conventional method which combines a time series clustering algorithm, namely K-Means with Dynamic Time Warping, with Complex Network Analysis. For the purpose of the present research, we have used an open database containing the data of 100 commercial and industrial consumers, collected every 5 minutes over a year. From the results, it is possible to identify different patterns of consumer behaviour and similar corporate profiles without any prior knowledge of the raw data.