Occupancy Forecasting for the Reduction of HVAC Energy Consumption in Smart Buildings

Buildings often operate under inefficient conditions and configurations due to their complex dynamics, a necessity of in-depth knowledge and intricate analysis tools. The fact is that interest in proposing higher order approaches for tackling efficiency problems in buildings has been steadily increasing during recent years. A wide range of approaches are being demonstrated, from model-predictive control schemes to frameworks for the detection of anomalies in energy consumption. Occupancy-centric methodologies, in particular, present one of the avenues with greatest potential of improvement because of their ability to adapt the behavior of the building to the real necessities of the users. This paper presents a novel occupancy modeling and forecasting methodology with the capability to support downstream demand-side management tools by providing accurate insight regarding the occupancy of spaces in the building. The proposed methodology takes advantage of the availability of presence detectors located on the different spaces of the building to study their dynamics and autonomously map their behavior. The complete methodology is validated experimentally in terms of accuracy and performance using real data from a research building.