An Approach of Reinforcement Learning Based Lighting Control for Demand Response

Abstract:
Lighting is a major contributor of building energy consumption. Lighting systems will thus be one of the important component systems of a smart grid for dynamic load management services such as demand response (DR). We consider the problem of autonomous control of multiple lighting systems in a building for providing DR Service, while keeping occupants’ illuminance comfort. To achieve an online and adaptive control for lightings, we propose to use reinforcement learning (RL) rather than other intelligent control algorithms to learn the lighting system environments with consideration of both DR signals and users’ illuminance requirements, for lighting control.