Model Free Optimal Control of Two Whole Buildings using Deep Q-networks

Abstract

This study introduces an application of deep Q-networks (DQN) for whole building energy optimization. The DQN was coupled to an office building and a retail building: an EnergyPlus simulation model provided by the U.S. Department of Energy (DOE). The optimal controls can successfully minimize total building energy use with appropriate indoor CO2 concentrations maintained (less than 1,000 ppm). It was shown that the DQN can find balanced control actions between different energy consumers in the building such as chillers, pumps, and air handling units (AHUs), leading to a whole building global optimization.