Uma aplicação em nuvem para comparar métodos de desagregação de energia através de uma base de dados industrial

Abstract

The deployment of smart meters makes it possible to collect energy consumption readings, where it is possible to extract detailed information about individual energy usage habits. The disaggregation process composes this scenario aiming at decomposing the total energy consumption measured at the household, commercial or industrial level into the contributions of individual electrical appliances. The use of disaggregated information, on the one hand, can be used to develop predictive models capable of predicting future energy consumption behaviors; on the other hand, it can be supplied directly to customers, so that an industry’s equipment has detailed energy use. Thus, an efficient consumption optimization is expected. This article presents an innovative proposal for a cloud architecture to deal with energy disaggregation in a non-intrusive way with real industrial data, using the methods of the Combinatorial Optimization (CO) and WaveNILM literature. The performance of the methods was compared using metrics known in the literature, such as F Score, MAE and NDE.