HVAC Cooling-Heating-Power Demand Strategy Optimization Based on DDPG Control

Electric energy and hot and cold energy are the basic energy needs of human life and the two main energy forms of integrated energy system. As a typical demand response resource, Heating Ventilation Air Conditioning (HVAC) can be introduced into the demand response problem of cooling-heating-power synergy. Aiming at the difficulties of traditional modeling such as self-learning ability and slow optimization speed, a model free deep reinforcement learning based HVAC cooling-heating-power demand optimization strategy was proposed. Considering the complex observation state and the prediction of continuous action space, a deep deterministic policy gradient (DDPG) was chosen to solve the demand response problem. Firstly, the thermal demand response model on the demand side was defined, and the HVAC environment model was established. Secondly, HVAC is described as a Markov decision process. Appropriate state actions, control signals, and reward functions are designed to speed up the optimization process. Finally, in order to verify the feasibility of the optimization strategy based on DDPG, the simulation results of cold and hot power under different constraints were introduced respectively. The results show that the typical HVAC system with multiple energy flows of cold and hot power has good demand-side response characteristics