Article poster: Occupancy Count Prediction for Model Predictive Control in Buildings

The concept of model predictive control (MPC) has been proposed as a method for optimizing energy consumption in buildings. MPC promises to deliver optimized building management without impeding indoor climatic properties. However, critical to the deployment of MPC are several factors such as weather forecasts and building occupancy predictions. In this poster, we focus on the latter and we present a method for predicting the number of people in buildings. The method relies on the availability of previous datasets of occupancy counts to accurately predict future occupancy counts in a building. In this poster we have utilized datasets from deployed 3D stereo-vision cameras in two rooms. We present the prediction accuracy of our method compared to both ground-truth data and the observed camera counts in the prediction period.