Modeling and Optimizing Building HVAC Energy Systems Using Deep Neural Networks

The heating, ventilating and air conditioning (HVAC) systems consumes more than half of the building energy consumption. An efficient HVAC system ensures the comfortable living and working environment. In this research, we proposes a novel approach to maximize the HVAC system efficiency regarding a typical office-type facility. One year energy data is collected from a commercial building in Chicago, IL for this research. The future room temperature and air humidity are predicted by deep neural networks. Deep neural network is also selected to extract the highly non-linear relationship between the control settings and the energy utility. A comparative analysis between the deep neural network and other state-of-arts data driven models is conducted. The deep neural networks outperforms the other algorithms and is applied to model the room temperature and air humidity. The predicted temperature and humidity are integrated into an energy optimization algorithm. Through numerical experiments, the systematic efficiency has been optimized to an acceptable level. Significant energy savings have been obtained while the facility's thermal comfort has been maintained.