Abstract
This paper discusses the adoption of a special Reinforcement Learning approach, namely the Actor-Critic (AC) one, to the temperature control of medium-size buildings. In particular, the approach here proposed is a predictive AC, in the sense that the AC algorithm is coupled not directly with the plant to be controlled, but with a predictor, thus enforcing a predictive behavior of the decision-making part of the AC. The reward function is defined on a prediction horizon, with 3 different formulas, as well as 3 different ways to compute the control action based on the predicted behavior. A real test case is considered and a comparison with a nonlinear MPC previously developed is given. Results are definitely encouraging in that predictive AC control reaches a similar performance as nonlinear MPC, without the need of realtime optimization.
Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
