Non-intrusive monitoring and identification of energy consumption of residential appliances

Author(s)

Rafael Cuerda Monzani

• Document type: Doctoral Thesis
• Press: Campinas, SP.
• Institution: Universidade Estadual de Campinas (UNICAMP). Faculdade de Engenharia Elétrica e de Computação
• Defense date: 2016-07-11

Examining board members

Walmir de Freitas Filho; Anna Diva Plasencia Lotufo; Jussara Farias Fardin; Fernanda Caseño Trindade Arioli; Luiz Carlos Pereira da Silva

• Advisor: Walmir de Freitas Filho

Abstract

Residential load monitoring provides useful information to specific entities such as government, utilities and consumers. Among these is the bill breakdown of the energy consumption pattern of each residence, enabling the energy estimation of individual appliances through load disaggregation techniques. In this thesis, real houses (in Brazil and abroad) were monitored and a computational tool to estimate the energy consumption of their major appliances was developed. The performance of the proposed algorithms was evaluated through a comparison against synthetic data generated by a residential network simulator, providing the home appliances¿ ON and OFF instants. Such tool enabled a comparison between the detected and classified events with those obtained by the proposed techniques. The data is acquired at the service entrance panel of each residence using a nonintrusive methodology, current and voltage data are locally processed and remotely sent to a server. Event detection algorithms are then run, combining data into a vector used as input to the load disaggregation procedure. Through this procedure, the events are associated with their specific loads, and then, the energy consumption is estimated based on their electrical signatures. The load disaggregation proposed by this thesis is performed automatically based on a general database, thus avoiding the initial training typical of other methodologies, such as neural networks. Moreover, this thesis proposes a tailored method to estimate the energy consumption of lighting using Monte Carlo simulation associated with the lighting time of use curve. The obtained results from the bill breakdown framework (detection, load disaggregation and energy estimation) of the analyzed residence and of the generated data by the residential network simulator are consistent and accurate (AU)

• FAPESP's process: 13/00437-0 - Non-intrusive monitoring and identification of energy consumption of residential appliances
• Grantee: Rafael Cuerda Monzani
• Support Opportunities: Scholarships in Brazil - Doctorate