jailcd2022038: A Review of Research on Reinforcement Learning-based Regulation Strategies for HVAC Control

Abstract

With the development of computer technology, artificial intelligence technology and automation control technology, model-free predictive control methods based on deep reinforcement learning have been developed relatively quickly. The main purpose of this technology is to meet the building load demand at the same time, it can help to realize the refinement of air conditioning system operation to improve the potential of air conditioning demand response. This paper classifies the reinforcement learning regulation methods into two main categories: value-based and strategy-based, and summarizes the advantages and disadvantages of various reinforcement learning methods and their applications on the sub-basis, and also provides an outlook on the reinforcement learning-based regulation technology for air conditioning systems