Integrating deep reinforcement learning into home energy management system based on soft actor-critic framework

Inhalt:
This paper proposes a two-layer home energy management model (HEMS) based on soft actor-critic (SAC) to deal with the uncertainty and complexity of family environment change. Firstly, the scheduling problem in home energy management is transformed into a Markov decision process, and the method of deep reinforcement learning is introduced to solve it. Then, the SAC method is introduced to solve the model. The first layer forms the economic scheduling of household schedulable load on the premise of ensuring comfort, and the second layer optimizes the charging and discharging process of energy storage device on the basis of the decision-making of the first layer, so as to further save the power cost. Finally, the household load public dataset REDD is used for verification. Experimental results show that the proposed method has better performance and computational efficiency.