Non-intrusive load monitoring (NILM) is a popular approach to estimate appliance-level electricity consumption from aggregate consumption data of households. Assessing the suitability of NILM algorithms to be used in real scenarios is however still cumbersome, mainly because there exists no standardized evaluation procedure for NILM algorithms and the availability of comprehensive electricity consumption data sets on which to run such a procedure is still limited. This paper contributes to the solution of this problem by: (1) outlining the key dimensions of the design space of NILM algorithms; (2) presenting a novel, comprehensive data set to evaluate the performance of NILM algorithms; (3) describing the design and implementation of a framework that significantly eases the evaluation of NILM algorithms using different data sets and parameter configurations; (4) demonstrating the use of the presented framework and data set through an extensive performance evaluation of four selected NILM algorithms. Both the presented data set and the evaluation framework are made publicly available.

Detecting when a household is occupied by its residents is fundamental to enable a number of home automation applications. Current systems for occupancy detection usually require the installation of dedicated sensors, like passive infrared sensors, magnetic reed switches or cameras. In this paper, we investigate the suitability of digital electricity meters -which are already available in millions of households worldwide -to be used as occupancy sensors. To this end, we have collected fine-grained electricity consumption data along with ground-truth occupancy information for 5 households during a period of about 8 months. Our results show that using common classification methods it is possible to achieve occupancy detection accuracies of more than 80%.

No abstract

Occupancy monitoring (i.e. sensing whether a building or room is currently occupied) is required by many building automation systems. An automatic heating system may, for example, use occupancy data to regulate the indoor temperature. Occupancy data is often obtained through dedicated hardware such as passive infrared sensors and magnetic reed switches. In this paper, we derive occupancy information from electric load curves measured by off-the-shelf smart electricity meters. Using the publicly available ECO dataset, we show that supervised machine learning algorithms can extract occupancy information with an accuracy between 83% and 94%. To this end we use a comprehensive feature set containing 35 features. Thereby we found that the inclusion of features that capture changes in the activation state of appliances provides the best occupancy detection accuracy.

No abstract