Profiling and Schedule Based Forecast of Electricity Usage to Improve Energy Performance of Administrative Buildings

Accurate forecast of electricity demand is an important problem in the energy domain. However, development of a cost-effective solution that does not require additional resources or data and makes hourly forecasts for several weeks, even a month remains an open challenge. Therefore, in this paper, we propose a method that allows to predict of electricity demand in administrative buildings relying on the concept of daily electricity load profiles. We identified different behavioral patterns during normal working days, COVID times, and vacations. By leveraging these patterns and historical data, we were able to generate more accurate predictions of electricity usage profiles than general forecasting techniques. We believe that this may help to develop a general approach which will provide a cost-effective solution that does not require additional resources or data. These ideas are demonstrated using the real data collected from administrative building located in Estonia.