Developing a Generalizable Model for Predicting Building Energy Usage

Nowdays, there are various methods for improving the energy efficiency of buildings, including the implementation of modern HVAC systems. However, evaluating measures aimed at improving energy efficiency can be a difficult task, since it is impossible to accurately measure how much energy a building would use without implementing any kind of energy-saving technology. Thus, an accurate quantitative assessment of mentioned effect has become possible because of advanced machine learning methods that are used for long-term energy consumption forecasting of buildings. Most of the research on related topics demonstrate non-representativity of the proposed models due to their narrow focus on a specific construction. This motivates the need of generalized models that can be applied to a wide range of building types, usage patterns, and consumption patterns. This paper presents a solution based on the analysis of physical regularities, which has a relatively high accuracy without reference to specific buildings.