Non-intrusive load monitoring (NILM) systems today represent a valid alternative to the installation of sensors dedicated to monitoring the main users powered by civil, commercial and industrial systems. As far as domestic use is concerned, these systems are based on algorithms capable of learning from the data collected on a certain number of houses, and then generalizing on houses never seen before. The training of these algorithms will be all the more efficient, the more consistent the training data is available. The aim of this work is the proposal of a modular architecture for a system that allows the development and testing of NILM systems. The system essentially allows the acquisition of training data in a flexible way, depending on the needs of the application, even for long periods (several weeks). The data are metrologically reliable, being acquired with a board certified as class 0.2, single-phase meter. In this work the NILM problem is introduced and the state of the art is briefly defined, with particular reference to Deep Learning (DL) systems. Finally, an architecture of a data collection system aimed at training and testing NILM algorithms is proposed.

Advanced Architecture for Training and Testing NILM Systems

Mari, S
;
Bucci, G;Ciancetta, F;Fiorucci, E;Fioravanti, A
2022

Abstract

Non-intrusive load monitoring (NILM) systems today represent a valid alternative to the installation of sensors dedicated to monitoring the main users powered by civil, commercial and industrial systems. As far as domestic use is concerned, these systems are based on algorithms capable of learning from the data collected on a certain number of houses, and then generalizing on houses never seen before. The training of these algorithms will be all the more efficient, the more consistent the training data is available. The aim of this work is the proposal of a modular architecture for a system that allows the development and testing of NILM systems. The system essentially allows the acquisition of training data in a flexible way, depending on the needs of the application, even for long periods (several weeks). The data are metrologically reliable, being acquired with a board certified as class 0.2, single-phase meter. In this work the NILM problem is introduced and the state of the art is briefly defined, with particular reference to Deep Learning (DL) systems. Finally, an architecture of a data collection system aimed at training and testing NILM algorithms is proposed.
978-1-6654-8360-5
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11697/194901
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus ND
  • ???jsp.display-item.citation.isi??? 0
social impact