Smart Meter Data Analytics for Occupancy Detection of Buildings with Renewable Energy Generation

Abstract

The activity of people causes distinctive patterns in the total electricity consumption of a building as does also the energy generation from renewables if present. The goal of the study is to test analysis methods that enable remotely metered electricity consumption data to be used for automatically analysing the occupancy of buildings equipped with renewable energy generation. Pattern recognition techniques can be used for the purpose of occupancy detection. It was concluded that in many cases the moving average of a consumption pattern is sufficient to determine the fact whether people occupy the building or not, but the addition of on-site renewable energy production makes it more complex. The fact if there is renewable energy production in the building plays a critical role and has to be detected additionally, because it generates fluctuations on its own which have to be taken into account. The averaging time step of the available datasets is another aspect that influences the precision of the achieved detection results. The occupancy information can be used for different purposes, like researching the mobility of humans. In this case the results need to be aggregated to preserve the privacy of individuals.