Residential Demand Response Strategy Based on Deep Deterministic Policy Gradient

With the continuous improvement of the power system and the deepening of electricity market reform, the trend of users' active participation in power distribution is more and more significant. Demand response has become the promising focus of smart grid research. Providing reasonable incentive strategies for power grid companies and demand response strategies for customers plays a crucial role in maximizing the benefits of different participants. To meet different expectations of multiple agents in the same environment, deep reinforcement learning was adopted. The generative model of residential demand response strategy under different incentive policies can be trained iteratively through real-time interactions with the environmental conditions. In this paper, a novel optimization model of residential demand response strategy, based on a deep deterministic policy gradient (DDPG) algorithm, was proposed. The proposed work was validated with the actual electricity consumption data of a certain area in China. The results showed that the DDPG model could optimize residential demand response strategy under certain incentive policies. In addition, the overall goal of peak load-cutting and valley filling can be achieved, which reflects promising prospects of the electricity market.

Automated summary

This study proposed an optimization model of residential demand response strategy based on the DDPG algorithm using deep reinforcement learning, which can be trained under different incentive policies to maximize the benefits of different participants. The results demonstrated good optimization performance of the model in practical applications, showing promising prospects for achieving the overall goal of the electricity market.