Architecture And Algorithm Design Of An Intelligent Home Energy Management System Based On Reinforcement Learning

The power industry is the foundation and lifeblood of China’s modern economy.In order to ensure the stable operation of the power system,smart grids have emerged as the times require.Facing the ever-changing power demand in the power market and the lack of building thermodynamic models,the Intelligent Home Energy Management System(IHEMS)realizes the goal of energy saving and environmental protection of the smart home while meeting the basic living needs of users through the integrated management of all smart devices in the home.Since there are different brands of smart devices on the market,an open IHEMS that can integrate smart devices of different brands can ensure that users can achieve optimal system integration on the basis of free choice of products.Compatible management of smart devices is a key issue that open IHEMS must solve.To this end,this paper designs an intelligent home energy management system architecture and intelligent home energy management algorithm based on reinforcement learning,which solves the problem of compatible management of smart devices and energy cost minimization of smart homes.The main technical features of the scheme proposed in this paper include:(1)Classify all smart devices and construct corresponding mathematical models according to specific device attributes that can be applied to energy management optimization.(2)Predict future real-time electricity prices,photovoltaic power and outdoor temperature data through the Bidirectional Gated Recurrent Units(Bi GRU)network,and use Discrete Wavelet Transform(DWT)to extract corresponding future prediction data Variation trends and local characteristics.(3)Model the smart home energy management problem using Markov decision process,and redesign the state space,action space and reward function,where the reward function is used to replace the objective function and constraints.(4)Based on the automatic entropy Soft Actor Critic(SAC)algorithm,an intelligent home energy management algorithm is designed to form a forecasting and decision-making integrated scheduling model to achieve optimal scheduling of various smart devices in the smart home.This paper simulates a smart home that includes smart devices such as photovoltaic panels,smart meters,HVAC,washing and drying machines,electric vehicles,and energy storage devices.Experimental results show that the proposed intelligent home energy management system architecture and intelligent home energy management algorithm can guarantee the basic living needs of users without a building thermodynamic model,significantly reduce the energy cost of smart homes,and can effectively cope with environmental changes caused by IHEMS scheduling performance degradation.