Short-term electricity consumption forecast with artificial neural networks — A case study of office buildings

To achieve climate and carbonization goals, electricity grid participants, such as buildings, must reduce their footprint trough renewable generation. Introducing storages can help buffering the fluctuating nature of renewable energy sources but only with future knowledge of consumption and generation, can batteries be scaled sensibly to economically viable options. An efficient energy management system and accurate energy forecasts are necessary to proactively work within battery limits, providing a short-term (day-ahead or hour-ahead) energy production plan which can then be utilized for demand response applications like load peak minimization, self-consumption optimization, intelligent energy storage, and predictive control. Focus of this paper is the accurate energy consumption prediction of office buildings and a case study, based on measurement data. The output of a prediction algorithm is intended to serve as input to predictive control models for a storage system, which enables efficiently managing energy storages and balancing consumption.