Deep Reinforcement Learning based Demand Response for Domestic Variable Volume Water Heater

Abstract

Domestic water heater is an important demand response resource in the home energy system. The variable volume water heater can dynamically adjust the water storage volume and participate in demand response for better energy efficiency. The system uncertainty caused by stochastic operating environments is an unavoidable challenge of the scheduling problem. Under this background, a deep reinforcement learning based optimization method is proposed to deal with the scheduling problem of the variable volume water heater, and the proposed method considers the comfort, safety and water hygiene of occupants. To achieve online automatic optimization, an optimization framework based on deep reinforcement learning method is established, and proposed an optimization algorithm to achieve cost-minimizing online scheduling. The simulation results show that the intelligent control method of variable volume water heater proposed in this paper can deal with the uncertainties of dynamic conditions, and the scheduled variable volume water heater can reduce energy cost by 22.7% than fixed volume water heater.